元素设计模式
Elemental Design Patterns
新泽西州上萨德尔河 • 波士顿 • 印第安纳波利斯 • 旧金山
纽约 • 多伦多 • 蒙特利尔 • 伦敦 • 慕尼黑 • 巴黎 • 马德里
开普敦 • 悉尼 • 东京 • 新加坡 • 墨西哥城
Upper Saddle River, NJ • Boston • Indianapolis • San Francisco
New York • Toronto • Montreal • London • Munich • Paris • Madrid
Capetown • Sydney • Tokyo • Singapore • Mexico City
制造商和销售商用来区分其产品的许多名称均已声明为商标。本书中出现这些名称时，如果出版商知道商标声明，则这些名称将以首字母大写或全部大写的形式印刷。
Many of the designations used by manufacturers and sellers to distinguish their products are claimed as trademarks. Where those designations appear in this book, and the publisher was aware of a trademark claim, the designations have been printed with initial capital letters or in all capitals.
作者和出版商在编写本书时已尽心尽力，但不作任何明示或暗示的保证，也不对错误或遗漏承担任何责任。对于因使用本文所含信息或程序而导致的或与之相关的偶然或间接损失，我们不承担任何责任。
The author and publisher have taken care in the preparation of this book, but make no expressed or implied warranty of any kind and assume no responsibility for errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of the use of the information or programs contained herein.
出版商在批量购买或特价销售时提供这本书的超值折扣，可能包括电子版和/或定制封面和内容，这些内容特定于您的业务、培训目标、营销重点和品牌兴趣。欲了解更多信息，请联系：
The publisher offers excellent discounts on this book when ordered in quantity for bulk purchases or special sales, which may include electronic versions and/or custom covers and content particular to your business, training goals, marketing focus, and branding interests. For more information, please contact:
美国企业和政府销售
(800) 382-3419
corpsales@pearsontechgroup.com
U.S. Corporate and Government Sales
(800) 382-3419
corpsales@pearsontechgroup.com
对于美国境外的销售，请联系：
For sales outside the United States, please contact:
International Sales
international@pearson.com
请访问我们的网站：informit.com/aw
Visit us on the Web: informit.com/aw
美国国会图书馆出版品目錄數據
Library of Congress Cataloging-in-Publication Data
Smith，Jason McC。（Jason McColm）
元素设计模式/Jason McC。Smith。p
. cm。
包括参考书目和索引。ISBN
0-321-71192-0（精装：alk. 纸）1. 软件模式。2. 软件架构
3. 系统设计。I. 标题。QA76.76.P37S657
2012
005.1—dc23
2012001271
Smith, Jason McC. (Jason McColm)
Elemental design patterns / Jason McC. Smith.
p. cm.
Includes bibliographical references and index.
ISBN 0-321-71192-0 (hardcover : alk. paper) 1. Software patterns. 2. Software architecture
3. System design. I. Title.
QA76.76.P37S657 2012
005.1—dc23
2012001271
版权所有 © 2012 Pearson Education, Inc.
Copyright © 2012 Pearson Education, Inc.
保留所有权利。印刷于美国。本出版物受版权保护，任何禁止复制、存储在检索系统中或以任何形式或任何手段（电子、机械、影印、录制或类似方式）传输之前，必须获得出版商的许可。要获得使用本作品材料的许可，请向 Pearson Education, Inc., Permissions Department, One Lake Street, Upper Saddle River, New Jersey 07458 提交书面申请，或将申请传真至 (201) 236-3290。
All rights reserved. Printed in the United States of America. This publication is protected by copyright, and permission must be obtained from the publisher prior to any prohibited reproduction, storage in a retrieval system, or transmission in any form or by any means, electronic, mechanical, photocopying, recording, or likewise. To obtain permission to use material from this work, please submit a written request to Pearson Education, Inc., Permissions Department, One Lake Street, Upper Saddle River, New Jersey 07458, or you may fax your request to (201) 236-3290.
ISBN 13：978-0-321-71192-2
ISBN 10：0-321-71192-0
ISBN 13: 978-0-321-71192-2
ISBN 10: 0-321-71192-0
文本由美国马萨诸塞州韦斯特福德的 Courier 公司使用再生纸印刷。
Text printed in the United States on recycled paper at Courier in Westford, Massachusetts.
2012 年 3 月第一次印刷
First printing, March, 2012
内容
Contents
1 Introduction to Design Patterns
1.2.1 Viewing Patterns as Rote
1.2.2 Language-Dependent Views
2.2 The Where, the Why, the How
2.2.1 Decomposition of Decorator
3.2.4 Stacked PINboxes and Multiplicity
4.5 Why You May Want to Read the Appendix
4.6.1 Focused Documentation and Training
6 Intermediate Pattern Compositions
7 Gang of Four Pattern Compositions
A.5 Composition and Reduction Rules
A.6 Pattern Instance Notation and Roles
A.7.11 Delegated Conglomeration
A.8 Intermediate Pattern Definitions
数字
Figures
2.1 Decorator’s usual example UML.
2.4 A simple method call as UML.
2.5 The parts of a method call.
2.7 A simple design space with EDPs.
2.9 The design space extended to three dimensions.
2.10 The design space with method similarity fixed to similar.
2.12 Deputized Redirection example UML.
3.1 UML collaboration diagram.
3.2 Strategy as pattern:role tags in UML.
3.3 Huge UML of a not-so-huge system.
3.4 UML 中以模式：角色标记的形式将策略作为多个实例。
3.4 Multiple instances of Strategy as pattern:role tags in UML.
3.6 Collapsed PINbox as annotation.
3.7 Singleton and Abstract Factory in class diagram.
3.8 Template Method in sequence diagram.
3.10 PIN used with UML class diagram.
3.11 PIN used with UML sequence diagram.
3.12 Standard PIN role connections.
3.13 Blank expanded PIN instance.
3.15 Expanded PIN instance using UML.
3.16 A need for multiple related PINboxes.
3.18 Multiple Strategy instances as PINboxes.
3.19 Showing the interaction between multiple Strategy PINboxes.
3.20 Abstract Factory as part of a larger UML diagram.
3.21 Abstract Factory subsumed within the expanded PINbox.
4.1 Abstract Interface and Inheritance EDPs as UML.
4.2 Fulfill Method的内部定义作为 UML。
4.2 Internal definition of Fulfill Method as UML.
4.3 Fulfill Method as simple connected PINboxes.
4.4 Fulfill Method as expanded PINbox.
4.5 Fulfill Method as standard PINbox.
4.6 在Fulfill 方法中翻转我们的 EDP —哎呀。
4.6 Flipping our EDPs in Fulfill Method—oops.
4.8 Alternative classes that can fulfill an Abstract Interface EDP.
4.9 Alternative structures that can fulfill an Inheritance EDP.
4.10 Decorator’s usual example UML.
4.11 Fulfill Method definition as annotated UML.
4.12 Objectifier UML annotated with PIN.
4.13 Objectifier and Trusted Redirection.
4.14 Object Recursion annotated with PIN.
4.15 Object Recursion as just PIN.
4.16 Object Recursion and Extend Method.
4.17 Decorator annotated with PIN.
4.19 Decorator instance as a PINbox.
4.20 Expanding Decorator: one level.
4.21 Expanding Decorator: two levels.
4.22 Expanding Decorator: three levels.
4.23 Expanding Decorator: four levels.
4.24 Delegation before Rename Method refactoring.
4.25 Delegation after Rename Method refactoring—Redirection.
4.26 Delegation before Move Method refactoring.
4.27 The design space with method similarity fixed to dissimilar.
4.28 Delegation after Move Method refactoring: boring case.
4.29 Delegation after Move Method refactoring: into same type.
4.30 Delegation after Move Method refactoring: Delegated Conglomeration.
4.31 Delegation after Move Method refactoring: Conglomeration.
4.32 Delegation after Move Method refactoring: Trusted Delegation.
4.33 Move Method重构后的委托：Revert Method。
4.33 Delegation after Move Method refactoring: Revert Method.
4.34 Delegation after Move Method refactoring: Deputized Delegation.
4.35 Summarizing refactoring effects so far.
4.36 Implicit used-by relationships among the EDPs and selected other patterns.
4.37 The full method-call EDP design space: dissimilar method.
4.38 The full method-call EDP design space: similar method.
4.39 Method-call EDP refactoring relations.
5.3 UI class cluster showing an instance of Trusted Delegation.
5.4 UI class cluster showing an instance of Trusted Redirection.
5.5 UI class cluster showing an instance of Deputized Delegation.
5.6 UI class cluster showing an instance of Deputized Redirection.
7.1 Abstract Factory subsumed within the expanded PINbox.
7.2 Reducing the diagram to just one instance of Abstract Factory.
7.4 Abstract Factory as PIN only.
7.5 Factory Method subsumed within the expanded PINbox.
7.6 Factory Method as PIN only.
7.7 Decorator subsumed with the expanded PINbox.
7.9 Decorator expanded three levels deep and flattened.
7.10 Proxy subsumed with the expanded PINbox.
7.12 Proxy PIN reorganized to better match Decorator.
7.13 Chain of Responsibility subsumed within the expanded PINbox.
7.14 Chain of Responsibility as PIN only.
7.15 Template Method subsumed within the expanded PINbox.
7.16 Template Method reduced to a single instance.
7.17 Template Method as PIN only.
7.18模板方法PIN 经过重新组织，以便更好地匹配装饰器。
7.18 Template Method PIN reorganized to better match Decorator.
7.19 Factory Method redefined using Template Method.
A.1 The full method call EDP design space: similar method.
A.2 The full method call EDP design space: dissimilar method.
表格
Tables
2.1 Pattern pieces sorted into three categories of a pattern definition
2.2 All interactions between entities of object-oriented programming
2.3 Nonscoping interactions between entities of object-oriented programming
A.1 All interactions between entities of object-oriented programming
A.2 Nonscoping interactions between entities of object-oriented programming
房源
Listings
2.1 A simple method call as pseudocode.
2.2 类、实例和命名空间内的字段，如 C++ 中定义和使用的。
2.2 Fields within classes, instances, and namespaces, as defined and used in C++.
2.3 A Java class, and one possible equivalent object and type.
2.5 A method call chain as pseudocode.
2.6 Simple method call for Figure 2.5.
2.7 Example of a Recursion method call in Java.
2.8 Example of a Delegation method call in C++.
2.9 Example of a Redirection method call in Objective-C.
2.10 Java 中Conglomeration方法调用的示例。
2.10 Example of a Conglomeration method call in Java.
5.4 Create Object Implementation.
5.6 Retrieve in a temporary variable.
5.7 Basic inheritance example in Objective-C.
5.8 Overriding an implementation.
5.9 Implementation assumption mismatch.
5.10 Obvious fix—but likely not feasible.
5.11 Fixing a bug while leaving old code in place.
5.12 Using Redirection to hide part of an interface.
5.13 Animals almost all move but in very different ways.
5.14 CEO delegates out responsibilities.
5.15 Tom paints the fence with help.
5.16 Prep work and cleanup are important.
5.17 Prep work and cleanup are decomposable.
5.18 Instance swapping for protocol fallback in C++.
5.19 Auto fallback/forward using Revert Method.
5.20 Using Redirection in Python to add behavior.
5.21 Using Extend Method to add behavior.
5.22 Inviting friends naively in Java.
5.23 A slightly better approach for inviting friends.
5.24 Delegated Conglomeration in Java.
5.25 Traditional iteration and invocation in C.
5.26 Object-oriented iteration and invocation in C++.
5.27 Basic Redirected Recursion in C++.
5.28 Paratroopers implementing Redirected Recursion.
5.29 UI widgets demonstrating Trusted Delegation in C++.
5.30 Event handler in C++ showing Trusted Redirection.
5.31 UI widgets demonstrating Deputized Delegation in C++.
5.32 UI widgets demonstrating Deputized Redirection in C++.
6.1 Conditionals to select behavior.
前言
Foreword
我突然想起了电影《2001：太空漫游》中一个精彩的场景。
There’s a wonderful scene in the movie 2001: A Space Odyssey that comes to mind.
大卫·鲍曼博士独自在废弃的发现号飞船上度过了几个月——此前，他还对流浪的哈尔进行了脑白质切除术——之后，他靠近一块巨石，这块巨石将他带入了一个新世界。他最后给地球发的信息是“这里满是星星！”
Having spent several months alone on the derelict ship Discovery—and that after having earlier lobotomized the errant Hal—Dr. David Bowman approaches a monolith that draws him in to a new world. His final message back to earth ends “It’s full of stars!”
软件密集型系统是我们用自己的脑力劳动创造的新世界。鲍曼看到的世界是由原子组成的，因此充满了星星，而我们的世界是由比特组成的……充满了图案。
Software-intensive systems are new worlds that we create with our own mental labor. Whereas the world that Bowman saw was formed from atoms and thus full of stars, our worlds are formed from bits...and are full of patterns.
无论有意还是无意，所有结构良好、软件密集型的系统都充满了模式。识别系统中的模式有助于提高对该系统进行推理的抽象程度；将模式强加于系统有助于为该系统带来进一步的秩序、优雅和简单性。根据我的经验，模式是过去二十年软件工程领域最重要的发展之一。
Whether intentional or not, all well-structured, software-intensive systems are full of patterns. Identifying the patterns in a system serves to raise the level of abstraction in reasoning about that system; imposing patterns on a system serves to bring even further order, elegance, and simplicity to that system. In my experience, patterns are one of the most important developments in software engineering in the past two decades.
我很高兴与 Jason 合作，他一直在改进 SPQR 的工作，我向你保证，他对模式的理解和实践做出了巨大贡献。《基本设计模式》将帮助你以一种新的方式思考模式，这种方式将帮助你应用模式来改善你创建和发展的软件世界。如果你是模式新手，这是一本很好的书，可以帮助你开始你的旅程；如果你是模式老手，那么我希望你会学到一些新东西。我确实做到了。
I’ve had the pleasure of working with Jason as he evolved his work on SPQR, and let me assure you that he has contributed greatly to the advance of the understanding and practice of patterns. Elemental Design Patterns will help you think about patterns in a new way, a way that will help you apply patterns to improve the software worlds that you create and evolve. If you are new to patterns, this is a great book to start your journey; if you are an old hand with patterns, then I expect you’ll learn some new things. I certainly did.
Grady Booch
IBM 研究员
2012 年 2 月
Grady Booch
IBM Fellow
February, 2012
前言
Preface
本书介绍了一种新的设计模式，即基本设计模式，它为软件工程设计模式的研究和应用奠定了基础。它的基础是对软件编程理论结构的研究，但它旨在实用和务实。它既适合初学者，也适合经验丰富的开发人员。它应该可以帮助学生参与软件行业，并为研究人员提供新的思考点。
This book is an introduction to a new class of design pattern, the Elemental Design Patterns, which form a foundation for the study and application of software engineering design patterns. Its foundations are in research into the very fabric of software programming theory, but it is intended to be practical and pragmatic. It is intended for both the beginning programmer and the seasoned developer. It should help students engage with the software industry and give researchers new points to ponder.
简而言之，这本书就是用来使用的。
In short, this book is meant to be used.
读完本章后，您将掌握一套新工具，对我们每天都在使用的编程基本概念有更深入的了解，并了解它们如何相互关联和交互以完成令人惊叹的工作。基本设计模式 (EDP) 是一组基本的编程思想，我们会不假思索地使用这些思想。本章提供了明确的描述、用于讨论的规范名称，以及一个用于协同使用和比较它们自身优缺点的框架。如果您是新生，您将了解到，您不必将不断增长的设计模式文献视为令人生畏的“全有或全无”模块的集合，而是有机会一点一点地学习它们，并逐渐有条不紊地理解这些文献。如果您是软件设计和模式方面的老手，您将找到新的方式来看待旧方法，并发现我们学科的新机遇。
By the end of it, you should have a new set of tools in your toolbelt, a richer understanding of some of the basic concepts of programming that we all use every day, and knowledge of how they relate and interact with one another to do amazing things. The Elemental Design Patterns, or EDPs, are a collection of fundamental programming ideas that we use reflexively and probably don’t think twice about when doing so. This body of work gives them explicit descriptions, regularized names to use in discussions, and a framework for using them in concert and for comparing them on their own merits. If you’re a new student, you’ll learn that instead of facing the ever-growing design patterns literature as a collection of daunting all-or-nothing blocks, you have a chance to take them on piece by piece and gradually understand the literature in a methodical way. If you’re an old hand at software design and patterns, you’ll find new ways to look at old approaches and see new opportunities for our discipline.
本书假设您对设计模式这一领域有一定了解，但尚未详细使用或研究过它们。知道它们的存在并对它们有一个简短的口语知识就足以开始讨论。本书并不假设您具有编程理论、语言设计的背景，甚至不假设您具有面向对象编程的强大背景，只是希望学习如何批判性地思考软件设计。这些主题将会涉及，但仅作为那些有兴趣通过以下方式深入研究它们的人的起点提供的参考资料。统一建模语言用于描述小例子，如果你还不了解 UML，我建议参考 [ 20 ] 或 [ 33 ]。你应该有编程的基本基础，无论是面向过程的还是面向对象的。后者会有所帮助，但不是绝对必要的——本文提供了大量必要的信息，以易于理解的块来解释面向对象编程。有面向对象系统经验的开发人员可能仍然会惊讶地发现，他们认为自己早就掌握了概念的新视角，以及对整个面向对象编程的更深理解。
This book assumes you have a passing familiarity with design patterns as a field but have not used or studied them in detail. Knowing that they exist and having a brief colloquial knowledge of what they are is enough to start the discussion. The book does not assume you have a background in programming theory, language design, or even a strong one in object-oriented programming, just a desire to learn how to think critically about software design. These subjects will be touched on but only as a starting point for those interested in diving deeper into them through the provided references. The Unified Modeling Language is used to describe small examples, and I suggest either [20] or [33] as references if you do not already know UML. You should have a basic foundation in programming, either procedural or object oriented. The latter will help, but it’s not absolutely required—this text provides much of the necessary information to explain object-oriented programming in easily digestible chunks. Developers experienced with object-oriented systems may still be surprised at finding new perspectives on concepts that they thought they had mastered long ago and a greater appreciation for object-oriented programming as a whole.
许多程序员认为“设计模式社区”是一个深奥的专家团体，他们自己并不是其中的一员。通过给你一个关于什么可以构成设计模式的新视角，本书应该会让你相信，每个程序员都是设计模式社区的成员，无论他们是否知道这一点。每个程序员在编写一行代码时都会使用设计模式，即使他们不这么认为。他们也不大可能意识到他们可以使用哪些选项。设计模式是共享的概念空间，我们在其中书写塑造我们世界的电子梦想。现在是时候绘制一张我们工作和娱乐的景观地图了。
Many programmers see the “design patterns community” as an esoteric body of experts and one that they themselves are not a part of. By giving you a new perspective on what can constitute a design pattern, this book should convince you that every programmer is a member of the design patterns community, whether they know it or not. Every single programmer uses design patterns every time they write a line of code, even if don’t think of it that way. Nor are they likely to realize the options they have at their disposal. Design patterns are the shared conceptual space in which we write the electronic dreams that shape our world. It’s time we had a map of the landscape in which we work and play.
本书以开创性的四人帮文本 [ 21 ] 为范例，分为两个部分。首先，讨论本书的写作原因和写作对象，并解释什么是 EDP、EDP 的来源以及 EDP 的重要性。本节解释了 EDP 背后的原理和原因。接下来，介绍模式实例符号，这是一种用于在多个粒度级别和多种环境中处理模式的图表系统。第一部分最后讨论了如何使用 EDP 来构建和结合更广泛的设计模式文献。本书的第二部分是设计模式的集合，从 EDP 开始，通过示例来说明它们如何组合形成中级模式，最后，选择四人帮模式重铸为 EDP 组合。此处介绍的 EDP 只是 EDP 目录的一部分，EDP 目录是第一轮定义和描述的基本模式的集合。随着基础概念的扎根，软件工程社区将继续定义和完善其他 EDP。我希望您决定为这一努力提供帮助。
Following the example of the seminal Gang of Four text [21], this book is divided into two sections. First is a discussion of why this book was written and who it is written for and an explanation of what EDPs are, where they came from, and why they’re important. This section explains the rationale, the why, behind the EDPs. Next is an introduction to the Pattern Instance Notation, a diagramming system for working with patterns at many levels of granularity and in a multitude of environments. Wrapping up this first section is a discussion of how EDPs can be used to build up to, and in conjunction with, the greater design patterns literature. The second section of the book is a collection of design patterns, starting with the EDPs and working through examples of how they combine to form Intermediate patterns, and finally, a selection of the Gang of Four patterns recast as EDP compositions. The EDPs presented here are only a portion of the EDP Catalog, a collection of the first round of defined and described fundamental patterns. The software engineering community will continue to define and refine additional EDPs as the underlying concepts take root. I hope you decide to help in the endeavor.
欢迎，很高兴您加入我们。
Welcome, it’s good to have you join us.
致谢
Acknowledgments
我要感谢很多人，感谢他们让这本书问世。以下按时间顺序排列……
I have many people to thank for this book coming to life. In not quite chronological order...
来自北卡罗来纳大学教堂山分校的 David Stotts，我的博士导师，多年来一直监督 SPQR 和 EDP 的诞生；还有我的委员会，尽管他们确信这可能是不可行的，但他们认为这将是一次有趣的旅程，并让我无论如何都去争取机会：Jan Prins、David Plaisted、Al Segars 和 Sid Chatterjee。你们每个人都在关键时刻提供了宝贵的帮助。
From the University of North Carolina at Chapel Hill, David Stotts, my Ph.D. advisor who oversaw the birth of SPQR and the EDPs over many years; also my committee, who, even though they were convinced it was probably infeasible, thought it would be an interesting journey and let me go for the brass ring anyway: Jan Prins, David Plaisted, Al Segars, and Sid Chatterjee. You each added invaluable help at critical times.
在我任职于纽约 IBM Watson Research 的那几年，我又一次遇到了 Sid Chatterjee，他说服我加入 Big Blue Playpen；还有 Clay Williams，他让我可以自由地进一步追求这些疯狂的想法，我仍然怀念和他一起喝咖啡；还有 Peter Santhanam，他支持这些想法，我从他身上学到了对遗留系统的更多欣赏；还有 Brent Hailpern，我从他身上学到了很多宝贵的管理经验，包括企业生活的黑色幽默和简单的人性；还有 Edith Schonberg，她比任何经理都更能忍受我的恶作剧；还有许多其他人，他们每次都听我疯狂地谈论这项工作。我的朋友们，我想念你们所有人。
From my years at IBM Watson Research in New York, Sid Chatterjee again, who convinced me to come play in the Big Blue Playpen; Clay Williams, who gave me free rein to pursue these crazy ideas further and with whom I still miss having coffee; Peter Santhanam, who championed those ideas and from whom I learned a greater appreciation for legacy systems; Brent Hailpern, from whom I learned many valuable lessons in management, the dark humor of corporate life and simple humanity; Edith Schonberg, who put up with my shenanigans more than any manager should have to; and many others who listened to me maniacally talk about this body of work at every turn. My friends, I miss you all.
同样来自 IBM 的，但值得特别提及的是 Grady Booch，他带我经历了一段我无论如何也不愿放弃的狂野旅程。Grady，你的指导、辅导和支持无可估量，我期待未来的合作和持续的友谊。
Also from IBM but deserving a special mention, Grady Booch, who took me under his wing for a wild ride that I wouldn’t have traded for anything. Grady, your guidance, mentoring, and advocacy have been immeasurable, and I look forward to future collaborations and continued friendship.
我现在在华盛顿州柯克兰的 Software Revolution, Inc. 担任高级计算机科学家，我要感谢大家对我把这些信息写成纸质文档的理解和支持。与你们所有人一起工作真的非常愉快，我迫切希望看到我们能把公司带向何方。
From The Software Revolution, Inc., in Kirkland, Washington, where I am now Senior Computer Scientist, I have to thank everyone for being understanding and supportive of my need to commit this information to paper. It has been a true pleasure working with all of you, and I am eager to see where we can take our company.
对于我的许多评论者来说，你们的建议和评论非常有见地，很有帮助。你们让这本书变得更好，我向你们表示最深切的感谢：Lee Ackerman、Lars Bishop、Robert Bogetti、Robert Couch、Bernard Farrell、Mary Lou Hines Fritts、Gail Murphy、Jeffrey Overbey、Ethan Roberts、Carlota Sage、Davie Sweis、Peri Tarr 和 Rebecca Wirfs-Brock。Addison-Wesley 的 Elizabeth Ryan、Raina Chrobak、Chris Zahn 和 Chris Guzikowski 在这一过程中给予了我们关爱和支持，在此我要感谢你们和那里的其他工作人员，特别感谢 Carol Lallier，她对本书的专业润色非常宝贵。
To my many reviewers, your advice and comments were highly insightful and helpful. You made this book a much better product, and you have my deepest thanks: Lee Ackerman, Lars Bishop, Robert Bogetti, Robert Couch, Bernard Farrell, Mary Lou Hines Fritts, Gail Murphy, Jeffrey Overbey, Ethan Roberts, Carlota Sage, Davie Sweis, Peri Tarr, and Rebecca Wirfs-Brock. Elizabeth Ryan, Raina Chrobak, Chris Zahn, and Chris Guzikowski at Addison-Wesley were the model of compassionate support during the trials of this process—my thanks to you and the rest of the crew there, with a special thanks to Carol Lallier, whose expert polish on this book was invaluable.
就我个人而言，我要感谢我的朋友和家人，在我为此投入了无数的时间和精力时，他们一直非常耐心，尽管他们希望在我搬回西雅图地区后能更多地了解我。
On a personal note, I thank my friends and family, who have been incredibly patient while I have put in seemingly endless hours on this, even though they were hoping to see more of me now that I’ve moved back to the Seattle area.
最后，我的妻子利亚。在我们在一起的这段时间里，你以各种大大小小的方式支持着我。你付出了时间、耐心和爱，我非常爱你，也非常感激你。谢谢你。语言无法表达我的感激之情。
Finally, my wife Leah. You have supported me in so many large and small ways throughout our time together. You have given your time, your patience, and your love, and you have my immense love and gratitude. Thank you. Words are simply inadequate.
谢谢大家。你们每个人都以某种方式为这些想法和本文的完善做出了贡献。这可能是我的孩子，但它有许多助产士。
Thank you all. Every one of you contributed in some way to the refinement of these ideas and this text. This may have been my baby, but it had many midwives.
— Jason McC. Smith
西雅图
2011 年 9 月 4 日
— Jason McC. Smith
Seattle
September 4, 2011
关于作者
About the Author
杰森·麦克。史密斯于 2005 年获得北卡罗来纳大学教堂山分校计算机科学博士学位，元素设计模式就是在该校模式查询和识别系统项目中诞生的。史密斯博士因在北卡罗来纳大学教堂山分校进行的研究而获得两项美国专利，一项是与 SPQR 相关的技术，另一项是 FaceTop 分布式文档协作系统。
Jason McC. Smith received his Ph.D. in computer science in 2005 from the University of North Carolina at Chapel Hill, where the Elemental Design Patterns were born as part of the System for Pattern Query and Recognition project. Dr. Smith has been awarded two U.S. patents for research performed at UNC-CH, one for technologies related to SPQR and one for the FaceTop distributed document collaboration system.
在此之前，史密斯博士在华盛顿大学获得物理学和数学双学士学位，在业界担任物理模拟工程师和顾问多年。他参与的项目包括声纳和海洋环境模拟、电子工程模拟、商用和军用飞机飞行模拟以及实时图形训练系统。
Prior to that, Dr. Smith spent many years in industry as a physics simulation engineer and consultant building off of dual B.Sc. degrees in physics and mathematics from the University of Washington. Projects of note included sonar and oceanic environment simulation, electronic engineering simulation, commercial and military aircraft flight simulation, and real-time graphical training systems.
在 IBM Watson Research 工作的四年为史密斯博士提供了一个机会，让他能够将 SPQR 和 EDP 目录以及组合方法的经验应用于大量传统和现代软件。
Four years at IBM Watson Research provided Dr. Smith with an opportunity to apply the lessons of SPQR and the EDP catalog and compositional approach to immense bodies of software, both legacy and modern.
史密斯博士目前是华盛顿州柯克兰市软件革命公司的高级研究科学家，他在那里继续完善 EDP 目录并寻找方法来增强公司对遗留系统进行自动化现代化和改造的目标。
Dr. Smith is currently Senior Research Scientist at The Software Revolution, Inc., in Kirkland, Washington, where he continues to refine the EDP catalog and look for ways to enhance the company’s goal of automated modernization and transformation of legacy systems.
1. 设计模式简介
1. Introduction to Design Patterns
无论以何种标准衡量，设计模式都是软件工程领域最成功的进步之一。然而，设计模式的历史却很奇怪，在发展过程中，它们最初的实用性和优雅性被遗忘、错放或只是传达错误。这本书可以为那些有设计模式经验的人填补一些可能的空白，并为刚接触该文献的学生提供一种更好的方式，让他们一点一点地吸收它。归根结底，设计模式文献本身就是一个相当大的信息块的集合，这些信息块的消化程度各不相同。这本书是一个基础，为熟悉设计模式的从业者提供了一种将这些信息块放入更大的理解系统中的方法，为刚接触设计模式的学生提供了一种从基本原理和单个有意义的小块中学习它们的方法。基本设计模式是真正的基本模式，因为它们构成了设计模式作为一门学科的基础。
Design patterns are one of the most successful advances in software engineering, by any measure. The history of design patterns is a strange one though, and somewhere along the way, much of their original utility and elegance has been forgotten, misplaced, or simply miscommunicated. This book can fill in some possible gaps for those who have experience with design patterns and can provide students new to the literature a better way of consuming it bite by bite. When it comes down to it, the design patterns literature as it stands is a collection of rather large nuggets of information of varying degrees of digestibility. This text is a foundation that provides a practitioner familiar with design patterns a methodology for placing those nuggets into a larger system of understanding and provides the student new to design patterns an approach for learning them from basic principles and in smaller pieces that make sense individually. The Elemental Design Patterns are truly elemental in that they form a foundation for design patterns as a discipline.
软件工程界的集体智慧是我们最宝贵的资产之一，我们还有很多东西需要相互学习。本书及其所基于的研究试图揭示我们在设计模式方面所失去的东西。在此过程中，它有助于实现最初的意图通过建立更好的模式共享讨论机制，我们能够更好地理解设计模式，让我们更深入地了解我们生产和使用的软件。我们的社区已经产生了广泛的设计模式，但我们缺乏深度。也就是说，我们对广泛的领域有广泛的了解，但将它们拼凑成一个综合整体的能力却很弱。这让我想起了从炼金术到现代化学的历史转变——在元素周期表出现之前，许多聪明研究人员的集体智慧是精确的，但没有很强的相关性。可以说，德米特里·门捷列夫的原始元素周期表的最大影响并不在于它为化学家提供了一种识别物质构成块之间模式的方法，而在于它提供了一种利用这些模式预测当时未被发现的元素性质的方法。镓和锗是第一个例子，门捷列夫在发现它们之前就准确地描述了它们的化学和物理性质。元素周期表将化学从描述性学科推进到预测科学。
The collective wisdom of the software engineering community is one of our most valuable assets, and we still have much to learn from each other. This book and the research on which it is based are an attempt to bring to light some of what we have lost regarding design patterns. In the process, it helps fulfill the original intent of design patterns by establishing a better mechanism for shared discussions of patterns, giving us a richer understanding of the software we produce and consume. Our community has produced a breadth of design patterns, but what we lack is depth. That is, we have a broad understanding of wide areas but only a weak ability to stitch them together into a comprehensive whole. It reminds me of the historical transition from alchemy to modern chemistry—until the periodic table came along, the collective wisdom of many intelligent researchers was precise but not strongly correlated. Arguably, the biggest impact of Dmitri Mendeleev’s original periodic table was not so much that it provided a way for chemists to identify patterns between the building blocks of matter but that it provided a way to use those patterns to predict properties of then-undiscovered elements. Gallium and germanium were the first examples of this, with Mendeleev accurately describing their chemical and physical properties well before their discovery. The periodic table advanced chemistry from descriptive discipline to predictive science.
设计模式在软件工程社区中的出现始于 1995 年出版的具有开创性的《设计模式：可重用面向对象软件的基础》。四人帮 (GoF)，Erich Gamma、Richard Helm、Ralph Johnson 和 John Vlissides，汇集了 Gamma 1991 年博士论文之后在研究和学术界流传的各种智慧。这项工作大量借鉴了 Christopher Alexander 在 20 世纪 60 年代的早期工作。Alexander 是一名土木工程师和建筑师，他的工作重点是寻找在特定环境中针对特定力量集的解决模式。他的主要见解是，建筑中有两种类型的设计，他称之为无意识的和有意识的。
The emergence of design patterns within the software engineering community began with the publication of the seminal Design Patterns: Elements of Reusable Object-Oriented Software in 1995. The Gang of Four (or GoF), Erich Gamma, Richard Helm, Ralph Johnson, and John Vlissides, gathered the various collected wisdoms that had been percolating in the research and academic communities following Gamma’s 1991 PhD dissertation. That work drew heavily from Christopher Alexander’s earlier work in the 1960s. Alexander was a civil engineer and architect, and his work focused on finding patterns of solutions to particular sets of forces within particular contexts. His primary insight was that there are two types of design that occur within architecture, what he named unselfconscious and selfconscious.
无意识的设计最常见于所谓的原始文化：每次都会忠实地复制房屋设计，并通过学徒制来确保对特定设计的忠诚和忠实。这种设计很少改变，坚持既定的形式被认为是目标，主要是因为特定的设计是几百年或几千年反复试验的结晶——它有效，如果它没有损坏，就不要修理它。虽然为人们提供住房的问题是普遍存在的，但发生该问题的各种环境，如雨、沙漠、冰、沼泽和森林，产生了各种各样的风格和设计，但在特定环境中，单一设计可能被视为“唯一的解决方案”，并且考虑到其特定的环境需求，它往往非常有效。然而，设计的应用几乎没有个人的判断或决策。
Unselfconscious design is most often seen in so-called primitive cultures: a house design is copied faithfully, every time, and apprenticeships are used to ensure fidelity and faithfulness to that particular design. This design changes rarely, and adherence to a given form is considered the goal, primarily because the particular design is the distillation of centuries or millennia of trial and error—it works, and if it ain’t broke, don’t fix it. While the problem of providing people with housing is universal, the various contexts in which that problem occurs, such as rain, desert, ice, swamp, and forest, give rise to an amazing array of styles and designs, but within a particular context, a single design may be considered “the only solution,” and it is frequently incredibly effective given its specific environmental needs. The design, however, is applied without much, if any, individual discretion or decision making.
自我意识设计是一种更现代的发明；设计师几乎可以在涉及风格、美学和材料的每一个环节自由地做出有意识的决定。这种建筑自由可以在各种现代建筑中看到在给定的地区、城市或街区内，房屋的风格和装饰风格各不相同。即使在您自己的街道上，您也可能会看到各种各样的风格和独特的装饰，每一种都是建筑师经过深思熟虑后做出的决定的结果。现代设计师有多种风格可供选择，一般来说，唯一的问题就是平衡房主的审美观和钱包。当然，这些房子满足了为居住者提供屋顶的基本标准，但当有如此多的其他轴线可以用来检验设计时，这个标准就太低了。当设计师可以自由地做任何事情时，从几乎无限的不合适或完全糟糕的解决方案中挑选出有效的解决方案就变得更加困难。建筑规范是我们试图限制住房设计中糟糕选择的一种方式，但即使以这些为起点，任务也是艰巨的。仅仅阅读建筑规范并遵守它们并不能产生有效的建筑作品。建筑规范是通用的，但好的建筑会考虑到每个细节层面的环境，从全球纬度和区域气候模式到当地土壤等级和特定地点的地形或植物。
Selfconscious design is a more modern invention; the designer is free to make conscious decisions at almost every turn involving style, aesthetics, and materials. This architectural freedom can be seen in the wide array of modern architecture within a given locale, city, or neighborhood. Even on your own street, you are likely to see a plethora of styles and distinctive flourishes, each the result of many conscious decisions on the part of the architect. The modern designer has a wide palette of styles to choose from, and generally speaking, the only problem is balancing the owner’s aesthetics and wallet. Sure, the houses meet the basic criteria of putting a roof over the occupants’ heads, but that’s a pretty low bar to set when there are so many other axes on which a design can be examined. When a designer is freed to do anything, it becomes even harder to pick the effective solutions out of the nearly infinite set of inappropriate or just plain bad ones. Building codes are one way we try to limit the bad choices in housing design, but even given those as a starting point, the task is daunting. Merely reading building codes and adhering to them is not going to produce an effective work of architecture. Building codes are generic, but good architecture takes into consideration the environment at every level of detail, from global latitude and regional weather patterns to local soil grades and site-specific terrain or foliage.
几乎在任何城镇或城市，你都能看到这种自觉设计的结果。一栋房子可能是乔治亚式的，一栋是仿维多利亚式的，另一栋是现代玻璃钢盒子式的，或者是错层式的，或者是牧场式的，或者是其他任何风格、种类和类型的建筑、材料和建筑。然而，我们必须扪心自问，这些设计是否是针对特定环境、特定背景的最佳解决方案，甚至只是有效的解决方案。例如，德克萨斯州奥斯汀可能不是建造无遮挡玻璃幕墙建筑的最佳地点，因为夏季阳光非常强烈，导致制冷成本大幅增加，从而产生额外费用。纽约州北部可能不适合建造平屋顶，因为隆冬时节数英尺厚的积雪会给房间梁增加很大的负荷。环境背景，即造成必须解决设计合适住房的一般问题的一系列力量，经常被忽视，而解决方案通常只能最低限度地令人满意，或者会产生必须解决的新问题。
You can see the results of this selfconscious design in almost any town or city. One house may be Georgian, one pseudo-Victorian, another a modern glass and steel box, or perhaps a split-level, a ranch, or any other number of styles, kinds, and types of construction, materials, and architectures. We have to ask ourselves, however, whether these designs work as optimal, or even just effective, solutions for that particular environment, for that particular context. Austin, Texas, for example, may not be the best place to build an unshaded glass-faced edifice because the sun is so intense in the summer, creating an added expense from the large increase in cooling costs. Upstate New York may not be an appropriate place for a flat roof because the weight of many feet of snow in the dead of winter adds a significant load to the room beams. The environmental context, the set of forces that create the situation in which the general problem of designing appropriate housing must be solved, is frequently ignored, and the solutions are generally only minimally satisfying or give rise to new problems that must be addressed.
显然，这适用于软件工程：我们几乎可以完成任何我们想到的事情，甚至比实体建筑的建筑师还要厉害。这是编程的惊人优势，也是它的致命弱点。我们几乎可以做任何事情，而且通常都能做到，但不幸的是，任何事情中好东西的子集非常小，我们的项目经常延期、超出预算，而且经常以惊人而悄无声息的方式失败。我们很少能带着成就感离开项目——更多的时候，我们觉得我们躲过了一劫。再次。为什么会这样？为什么，当我们拥有数十年的集体经验，以及可能数百万人年的工作经验时在实际工作中，我们是否每次遇到问题时都还在苦苦挣扎？一些设计师和开发人员似乎能够出色地避开复杂性，找到设计中有效的核心。我们其他人似乎永远被困在“因为我被告知要这样做”的无意识和自我意识设计的麻痹之间。
It should be apparent how this applies to software engineering: we are capable of doing nearly anything that pops into our heads, even more so than the architects of physical buildings. This is the amazing strength of programming—and its Achilles heel. We can do just about anything, and usually manage to do so, but unfortunately, the subset of good things out of the set of anything is quite small, and our projects are often late, over budget, and frequently fail in ways spectacular and quiet. Rarely do we walk away from projects with a feeling of accomplishment—more often, we feel we dodged a bullet. Again. Why is this? Why, when we have decades of collective experience, and quite possibly millions of tallied person-years in the field, are we still thrashing in the weeds every time we approach a problem? Some designers and developers seem to be phenomenally able to sidestep the complexity and find the kernel of effectiveness in a design. The rest of us seem to be perpetually stuck between the unselfconsciousness of “because I was told to” and the paralysis of selfconscious design.
亚历山大的作品试图减轻建筑师的这一问题，揭示原始文化在设计方面的有效性与现代建筑近乎痉挛的尝试一切的方法之间的差异。两者之间需要取得平衡。我们需要找到无意识建筑中存在的基本原则和一般解决方案，并以一种让它们有意识地、有意图地适用于各种环境的方式描述它们。通过反复试验来努力获得的各种解决方案的智慧需要提炼成一套概念，任何人都可以学习，应用于许多地方，并用作思考设计的指南。
Alexander’s work was an attempt to alleviate this problem for architects of buildings, to bring to light the disparity between the effectiveness of the primitive cultures at design and the nearly spastic try-anything approach of modern architecture. Somewhere in between is a balance to be struck. We need to find the underlying principles and general solutions that exist in unselfconscious architecture and describe them in a way that makes them applicable in a wide variety of contexts selfconsciously and with deliberate intent. The wisdom of the various attempts at solutions, hard-earned through trial and error, need to be distilled into a body of concepts that can be learned by anyone, applied in numerous places, and used as a guide for thinking about design.
这就是设计模式——一个充满活力和活力的社区对专业知识的提炼。这是众包的最佳表现。自最初的 GoF 工作以来，模式社区已经发展了十多年，规模庞大，充满活力，我们的产出量巨大。Grady Booch 和 Celso Gonzalez 一直在他们的网站上收集他们在工业界和学术界能找到的所有模式 [ 11 ]。到目前为止，他们已经拥有超过 2,000 个模式。这个社区的产出量非常庞大，尽管对质量有一些讨论，但更紧迫的问题还是规模问题。
This is what design patterns are—the distillation of expertise by an exuberant and robust community. This is crowdsourcing at its best. The patterns community that has grown over the decade-plus since the original GoF work is large and energetic, and our output is voluminous. Grady Booch and Celso Gonzalez have been collecting every pattern they can find in industry and academia at their website [11]. So far, they have over 2,000 of them. The quantity of output in this community is huge, and although there are some discussions about the quality, the more pressing problem is one of scale.
即使拥有一个完全索引、精心策划的高质量设计模式集合，对于非学者来说，信息量也实在太大，无法准确、快速地筛选。更糟糕的是，对于希望学习良好设计背后原则的学生来说，仅从良好设计的例子中学习是极其困难的。这有点像试图从跑道上的飞机检查中学习航空流动的数学知识。对于认为自己发现了新设计模式的经验丰富的模式从业者来说，没有现成的方法将新模式与现有模式进行比较，以了解其与既定文献的关系，也没有办法创建工具来支持这一需求。
Even with a fully indexed, well-curated collection of quality design patterns, there is simply too much information for a nonscholar to sift through accurately and quickly. Worse, it is incredibly difficult for a student wishing to learn the principles behind good design to do so solely from examples of good design. It is a bit like trying to learn the mathematics of aeronautical flow from inspecting aircraft on a runway. For experienced patterns practitioners who believe they have uncovered a new design pattern, there’s no ready way to compare a new pattern against existing patterns to see how it relates to the established literature, and there’s no way to create tooling to support this need.
软件开发社区需要的是对其现有资源的更透彻理解，一种解释如何更精确地描述现有设计模式的方法，并使用学生或新开发人员可以理解的组件和明确定义的原则来实现这一点。我们需要的是我们设计模式文献中的基本原则的指南以便我们能够更好地理解、教授和学习我们确定的最佳实践。本书是该指南的基础。
What the software development community needs is a more thorough understanding of what it has at its disposal, a methodology that explains how to more precisely describe the existing design patterns and does so using components and well-defined principles that are accessible to the student or new developer. What we need is a guide to the underlying basic principles of our design patterns literature so that we can better comprehend, teach, and learn our identified best practices. This book is a foundation for that guide.
1.1. 部落沉思
1.1. Tribal Musings
记录和传播已知的最佳实践所带来的效率很重要，但我们必须这样做的原因在我们的社区中却被很大程度上忽视了。坦率地说，我们终有一死，而我们这个年轻的领域正在衰老。我们已经失去了许多为我们的行业奠定基础的杰出人物，还有更多的人将很快离去。这只是生活中的一个事实，我们作为一门学科还没有做好充分的准备来应对它。
The efficiencies we gain from documenting and passing along known best practices are important, but the reason we must do so has been largely ignored in our community. To put it bluntly, we are mortal, and our young field is aging. Already we have lost a number of luminaries who established the groundwork for our industry, and many more will be gone soon. It is just a fact of life, one that we are poorly prepared to deal with as a discipline.
更糟糕的是，软件有一个特殊的特性，那就是寿命会远远超过预期寿命。COBOL 仍然是全球商业系统中一股不容小觑的力量。Fortran 仍然在世界科学建模软件中执行大量计算。目前，主要的高性能计算机系统在其固件中嵌入了三十多年前首次用汇编语言或 C 语言创建的代码。您几乎可以肯定，在您最近购买的个人计算机附带的数百万行实现中，有一段源代码目前无人能理解。
Worse still, software has a peculiar trait of living long past its expected lifetime. COBOL is still a force to be reckoned with in business systems around the globe. Fortran still performs much of the computation in the world’s scientific modeling software. Currently shipping major high-performance computer systems have code embedded deep in their firmware that was first created three decades or more ago, in assembler or C. You can be almost certain that somewhere in the millions of lines of implementation that came with your latest personal computer acquisition lies a piece of source code that no person currently understands.
我们知道应该记录我们的软件；我们知道应该保持其最新状态；我们知道应该致力于记录或筛选为什么、如何以及理由；但我们也知道这很麻烦。确实如此，所以我们不这样做。相反，我们拥有的是知识体系，它被锁在开发人员的头脑中，在需要时断断续续地传递，通常没有任何利益相关者之间共同理解的综合框架。
We know we should document our software; we know we should keep it up to date; we know we should commit to pen or screen the whys, the hows, and the reasons; but we also know it is a pain. It really is, so we don’t do it. What we have instead is a body of knowledge that is locked within the heads of developers, that is passed along in fits and spurts, when prompted and only where necessary, frequently without any comprehensive framework of common understanding among stakeholders.
Grady Booch 曾将“部落知识”一词广为人知，用来描述此类信息 [ 10 ]，这个词语非常适合用来描述此类信息。但它也有一些令人不安的推论。
Grady Booch has popularized the phrase “tribal knowledge” for such information [10], and it fits all too well. It also has some rather unsettling corollaries.
仅依靠口头传统传递知识的文化在带宽和准确性方面都受到限制，而且前提是它们具有传递信息的强大传统。在信息传递方面缺乏真实性和精确性的文化更容易受到腐败的影响。然而，强大的口头传统可以带来截然不同的结果。
Cultures that rely solely on oral tradition for the passing of knowledge are limited in both bandwidth and accuracy, and that’s assuming they have a strong tradition of passing along the information. Cultures with a weak discipline for veracity and precision in information transfer leave themselves open to more rapid corruption. A strong oral tradition, however, can result in a very different outcome.
开发社区的信息传递最终还是以口头形式进行。虽然我们可能会将自己理解的零碎内容写下来，但我们通常不会将自己的理解全部写下来，而且我们也不会将这种记录与系统的发展保持同步。这文档腐烂现象普遍存在，只有通过询问更多信息，我们才有可能填补空白，找出特定系统为何如此。
The development community has what is ultimately an oral tradition of information transfer. Although we may write down bits and pieces of what we understand, we frequently do not write down the entirety of our comprehension, and we do not keep such documentation in sync with the evolution of our systems. This document rot is pervasive, and only by asking around for further information can we hope to fill in the gaps to find out why a particular system is how it is.
说实话，这并不总是一件坏事。敏捷软件开发方法更倾向于工作代码而不是文档化代码，这种观点很难反驳。当然，除非它变得重要。敏捷系统有一种有趣的方式成为遗留系统，成长为成熟的代码库，需要更大的团队协同工作。最终，以敏捷工作开始的代码，如果成功的话，将面临与传统开发系统相同的许多挑战。开发人员离开。文档腐烂。知识丢失。
This isn’t always seen as a bad thing, to be honest. Agile software development methodologies prefer working code over documented code, and it’s hard to argue with this viewpoint. Until it matters, of course. Agile systems have a funny way of becoming legacy systems, of growing into mature codebases with larger teams that must work in concert. Eventually, code that started as an agile effort, if it is successful, will face many of the same challenges as traditionally developed systems. Developers leave. Documentation rots. Knowledge is lost.
目前的软件还不能准确地称为自文档化，而对于自动化系统来说，直接从源代码中提取以特定方式完成某件事的突出原因几乎是不可能的。这很不幸，因为我们想鱼与熊掌兼得。我们希望在需要时随时随地获得最新的文档，但我们不想被文档所累。我们希望我们的代码更加自文档化，或者至少可以自动文档化，但我们大多数人没有这种奢侈。所以我们只能望洋兴叹，希望一切顺利。与此同时，我们对系统的集体理解却在退化。最后，我们所拥有的最好描述就是非常薄弱的口头传统。
Software as it currently stands is not what anyone could accurately call self-documenting, and extracting the salient reasons why a thing was done in a particular way, directly from the source code, has been considered nearly impossible for an automated system. This is unfortunate, because we would like to have our cake and eat it too. We want up-to-date documentation when and where we need it, but we don’t want to be burdened with it otherwise. We’d like our code to be much more self-documenting, or at least automatically documentable, but most of us don’t have that luxury. So we punt and hope for the best. Meanwhile, our collective understanding of the system degrades. In the end, what we have is best described as a very weak oral tradition.
结果是，收集到的部落知识沦为“部落神话”。“为什么？”这个问题再也无法回答，只能说“因为我们一直都是这么做的。”我有一种隐隐的怀疑，如果你曾经是开发团队的新员工，你现在肯定会惊恐地点头。你在现实生活中讨论过这个问题，可能次数多得你都记不清了。
The result is that the collected tribal knowledge degrades into “tribal mythology.” “Why?” is not a question that can be answered any longer, except to say, “Because we’ve always done it that way.” I have a sneaking suspicion that if you have ever been the new hire on a development team, you’re nodding in horror right now. You’ve had that discussion in real life, probably more times than you care to recall.
部落神话是没有理解的行为。它是死记硬背的，没有任何基础来说明原因。部落神话在群体中活跃的其他迹象包括：“因为我就是这么学的。”“我不太确定，但乔说就是这样做的。”“简本可以告诉你的，但她去年退休了，所以只要抄袭就行了。”“哦不，不要改变它！它会坏掉，我们无法修复它。”这些评论表明未能理解行动背后的原因，或者至少不愿或无法将理解传递给听众。随着时间的推移，这种缺乏理解会滋生大量的不确定性和对变化的恐惧。不幸的是，在某种程度上，这是大多数项目的现状，具有讽刺意味的是，我们的行业是由创新、变革和最先进的技术进步驱动的。
Tribal mythology is action without comprehension. It is rote without any foundation on which to state why. Other indications that tribal mythology is active in a group include the following: “Because that’s how I was taught it.” “I’m not really sure, but Joe says that’s how its done.” “Jane could have told you, but she retired last year, so just copy what’s there.” “Oh no, don’t change that! It’ll break and we won’t be able to fix it.” These comments exhibit a failure to comprehend the reasons behind an action, or at least an unwillingness or inability to pass the comprehension along to the listener. Over time, this lack of understanding breeds a great deal of uncertainty and fear of change. Unfortunately, it is at some level the status quo on most projects, which is ironic given that our industry is driven by innovation, change, and advancement of the state of the art.
然而，部落智慧是这种部落神话的道德反面。它规定了行动，理解了行动，如何行动，为什么行动，并且具有适应性适应新环境、新情况和新问题。它超越了死记硬背，通过全面讨论行动背后的原因来提供启发。在过去的某个时候，对于系统中的每个行动或决定，几乎有人知道为什么这样做。将这些决定，即使是小决定，继续推进也是至关重要的。小决定会积累成大系统，小设计会构建成大设计。通过确保我们拥有促进理解的强大知识保留传统，我们建立了部落智慧的传统。
Tribal wisdom, however, is the virtuous flip side of this tribal mythology. It is prescribed action with understanding, how accompanied by why, and is adaptable to new environments, new situations, and new problems. It transcends rote copying, and provides illumination through a comprehensive discussion of the reasons behind the action. At some point in the past, for almost every action or decision in a system, someone knew why it was done that way. Carrying those decisions forward, even the small ones, can be critical. Small decisions accrete into large systems, and small designs build into large designs. By ensuring that we have a strong tradition of knowledge retention that facilitates understanding, we build a tradition of tribal wisdom.
设计模式旨在收集部落智慧。遗憾的是，它们经常被（错误地）当作部落神话，只应用“如何”而不理解“为什么”。
Tribal wisdom is what design patterns were intended to collect. Sadly, they are frequently (mis)treated as tribal mythology, by applying the how without a clear comprehension of the why.
如果你在职业生涯中还没有遇到过这种情况，让我再举一个可能对你有所启发的例子。最近，我和妻子买了第一套房子，还有第一个院子。我们居住的地区以多雨而闻名，因此也以苔藓而闻名。现在，我喜欢苔藓。它是绿色的，几乎不需要维护，而且是一种柔软的地被植物。它满足了院子的所有常规要求，而且比草坪所需的工作量要少，但我们遇到了一个奇怪的情况。院子的一部分被浓重的阴影遮蔽，很少甚至从未见过阳光。这个区域基本上是坚实的苔藓，没有草或任何其他植被。即使是耐阴的草也无法获得足够的光照来茁壮成长。
If you haven’t yet had the pleasure of running into this situation in your career, let me offer another example that may be illuminating. Recently my wife and I bought our first house, and with it, our first yard. The region we live in is renowned for its rain and consequently its moss. Now, I like moss. It’s green, it takes about zero maintenance, and it makes a nice soft ground cover. It satisfies all the usual requirements for a yard, with less work than grass requires, but we had an odd situation. Part of the yard is heavily shaded and rarely, if ever, sees sun. This area is basically solid moss, with no grass or any other vegetation. Even shade-tolerant grasses can’t get enough light to thrive.
然而，在阴暗部分两侧二十英尺的地方，大多数有太阳的日子里，阳光都能照到。苔藓在这一区域成片生长，但根据我最初的评估，我认为苔藓和草很好。苔藓和草很好地共存，苔藓并没有扼杀草，只是填补了草不太茂密的地方。在阳光最充足的地方，几乎没有苔藓，但有很多草。在阴暗的地方，有坚实的苔藓，但没有草。在过渡区域，两者共存。还有什么比这更好的呢？
Twenty feet on either side of the heavily shaded portion, however, sunlight is available on most days when the sun is actually out. Moss grows in patches through this section, but in my initial assessment, I thought it was fine. The moss and grass were coexisting nicely, and the moss wasn’t choking out the grass, merely filling in the places where the grass wasn’t quite so thick. In the sunniest areas, there was almost no moss but lots of grass. In the shadiest areas, there was solid moss but no grass. In the transitional regions, the two coexisted. What could be better?
不幸的是，看到这种情况的长期居民都惊慌失措。“你必须把所有的青苔都除掉！”当我问为什么时，我得到的答案是“因为它不是草。”“因为这是你要做的。 ”“因为它对草坪有害。”没有人能让我满意地告诉我为什么我应该除掉青苔。在我看来，如果我把所有区域的青苔都除掉，不管当地的微环境如何，我都会得到一个光秃秃的地方，草不会在阴凉处生长。这不是最理想的。
Unfortunately, long-time residents who saw this situation were horrified. “You have to get rid of all the moss!” When I asked why, I was met with answers such as “Because it’s not grass.” “Because it’s what you do.” “Because it’s bad for the lawn.” No one could tell me, to my satisfaction, why I should get rid of the moss. It seemed to me that if I removed all of the moss, in all areas, regardless of the local micro-environment, I would have a bare spot where grass wouldn’t grow in the shade. This was less than optimal.
更糟糕的是，就像许多软件项目的情况一样，我接手的情况是，我不知道以前的住户对院子做了什么维护，也不知道为什么。没有文件说明我应该如何打理草坪，或者为什么院子保持这种状态。所以我运行了几个实验。在最阴凉的地方，我拔出一小块苔藓，并在其中播下草的种子。在院子的其余部分，我让苔藓生长，看看会发生什么。
To make matters worse, as is the case in many software projects, I had inherited a situation in which I had no idea what the previous residents had done for maintenance in the yard or why. There was no documentation to indicate what I should do for my lawn or why the yard had been left in this configuration. So I ran a couple of experiments. In the shadiest areas, I pulled up a small section of moss and seeded it with grass. In the rest of the yard, I let the moss go to see what would happen.
最阴凉处的草籽从不茂盛。有些发芽了，但永远无法很好地扎根。如果按照部落神话的描述，我院子里的大部分地方都会是裸露的泥土，坦率地说，我更喜欢苔藓。苔藓郁郁葱葱，在那个区域生长茂盛，无需维护。对于那个特定区域，苔藓是一种很好的地面覆盖解决方案。
The grass seed in the shadiest area never thrived. Some sprouted, but it could never get established well. Applying the tribal mythology would have resulted in bare dirt in a good section of my yard, and frankly, I prefer the moss to that. It is green and lush, and it thrives in that area without maintenance. For that specific area, moss is a good ground cover solution.
在院子里其他没有树荫的地方，苔藓和草或多或少可以很好地共存。草长得很好，而苔藓无法直接克服它。不幸的是，苔藓有副作用。在阳光最充足的地方，苔藓可以作为杂草在下面发芽的保护层，防止鸟类和老鼠吃掉种子或嫩芽，并且保持适当的湿度。苔藓不会扼杀草，但杂草肯定会。通过让苔藓存在于阳光充足的地方，我为杂草提供了一个苗圃，当它们穿过苔藓时，它们在阳光下茁壮成长并迅速蔓延。苔藓还为杂草的根部提供了保护性的湿润层，为它们提供了一条畅通无阻的生长通道。
In the rest of the yard with little or no shade, it turns out that moss and grass do play well together, more or less. The grass grows nicely, and the moss can’t overcome it directly. Unfortunately, the moss has a side effect. In the sunniest areas, the moss acts as a protective layer for weeds to sprout underneath, safe from birds and mice who would eat the seeds or shoots, and properly moist. The moss won’t choke out the grass, but the weeds definitely will. By letting the moss exist in the sunny areas, I was giving weeds a nursery to get established, and when they penetrated the moss, they thrived in the sunlight and spread rapidly. The moss also provided a protective moist layer for the roots of weeds to travel along, offering them an unhindered growth channel.
在最阴暗的地方，这不是什么问题，因为那里根本没有足够的阳光让草或杂草生长良好，但在阳光充足的地方，情况就很糟糕了。几个月内，我与杂草展开了一场真正的地盘之战。苔藓本身从来都不是问题，但它为更大的问题埋下了伏笔。
In the shadiest areas, this wasn’t a problem, because there simply isn’t enough sun for grass or weeds to grow well, but in the sunny areas, it was horrible. Within a couple of months, I was fighting a literal turf war with the weeds. The moss was never a problem by itself, but it set the scene for a much larger one.
现在我知道了为什么要从草坪上除掉青苔。问题并不在于青苔，而在于青苔会形成次生微气候，从而引发严重的问题。从本质上讲，青苔会产生一组新的力量，形成新的环境。在新的环境、新的环境中，会出现新的问题，比如杂草。现在，除掉青苔的建议是有道理的，至少对于杂草成问题的区域来说是这样，比如我院子里阳光充足的地方。
And now I know the why behind removing moss from a lawn. It’s not so much the moss that is the issue, it is that the moss creates a secondary microclimate that sets up a serious situation. Essentially, the moss creates a new set of forces at play that form a new context. Within that new context, that new environment, new problems arise—like weeds. Now the advice to remove the moss makes sense, at least for areas where weeds are an issue, such as the sunny areas of my yard.
因为我知道原因，所以我现在可以根据环境力量调整我对这些知识的应用。在阳光充足的地方，我必须去除并防止青苔，这样杂草就不会成为以后的问题。在阴凉的地方，我选择不这样做，因为这样做会给我带来另一个问题，让我的土地变得光秃秃的，很难让草长得很好。
Because I know the why, I can now alter my application of this knowledge according to the environmental forces. In the sunny areas, I must remove and prevent the moss so that weeds are not a later problem. In the shady areas, I choose not to because doing so would create another problem for me, leaving me with bare dirt where I’d struggle to get grass to grow well.
事实上，因为我知道这些建议背后的原因，所以我可以根据具体情况（阳光充足还是阴暗）定制我得到的解决方案——“除掉青苔”，这不仅解决了我最初的问题，而且防止了新问题的产生。最初的部落神话现在变成了部落智慧，可以在适当的时间和地点分享、调整和应用。从本质上讲，这是一种模式的开始。
As it is, because I know the reasons behind the advice, I can custom fit the solution I was given—“Remove the moss”—based on the context—sunny vs. shady—and not only solve my initial problem but prevent new ones from being created. What was initially tribal mythology is now tribal wisdom that can be shared, adjusted, and applied when and where appropriate. In essence, it is the beginning of a pattern.
1.2. 艺术还是科学？
1.2. Art or Science?
毫无疑问，模式是一个蓬勃发展的模因，并且具有极大的实用性。现在整个学术会议都致力于模式，Ackerman 和 Gonzalez 的基于模式的工程正在成为一门独立的学科 [ 2 ]，行业顾问现在需要掌握这些知识并能当场绘制出它们的统一建模语言 (UML) 图。现有的工具可以生成、显示、生成和提取模式。模式作为一个整体，是软件工程领域的假定组成部分。我们只是不太确定它们如何融入该领域或如何相互适应。有两个问题阻碍了对模式进行更全面的方法研究，不幸的是，这些问题在行业中无处不在。第一个问题是将模式视为可以盲目复制的冻结元素，第二个问题是将特定于语言的模式实现与模式本身的变体相混淆。
There is no doubt that patterns are a thriving meme, and one with great utility. Entire academic conferences are now dedicated to patterns, Ackerman and Gonzalez’s patterns-based engineering is becoming a defined discipline in its own right [2], and industry consultants are now expected to have them under their belt and be able to whip out Unified Modeling Language (UML) diagrams of them on the spot. Tools exist to produce, display, generate, and extract patterns. Patterns, as a collective whole, are an assumed component of the software engineering landscape. We’re just not quite sure how they fit into that landscape or how they fit with each other. Two issues prevent a more comprehensive approach to patterns, and unfortunately they are ubiquitous in the industry. The first is treating patterns as frozen elements to be blindly copied, the second is confusing language-specific pattern implementations with variants of the patterns themselves.
1.2.1. 将模式视为死记硬背
1.2.1. Viewing Patterns as Rote
让十几位开发人员定义设计模式，您可能会得到十几个答案。在更传统的“特定环境中反复出现的问题的解决方案”答案中，您还可能听到诸如“配方”或“示例结构”或“一些示例代码”之类的短语，这暴露了对模式所提供的内容的狭隘看法。模式旨在进行变异、扭曲和塑造，以满足问题环境中特定力量的需求，但开发人员往往只是从模式文本或站点复制和粘贴示例代码，并宣称结果是模式的成功应用。这通常是失败的秘诀，而不是产生良好设计的秘诀。
Ask a dozen developers to define design patterns, and you’ll likely get a dozen answers. Among the more traditional “a solution to a recurring problem within a particular context” answers, you’re also likely to hear phrases such as “a recipe” or “an example structure” or “some sample code,” betraying a rather narrow view of what patterns provide. Patterns are intended to be mutated, to be warped and molded, to meet the needs of the particular forces at play in the context of the problem, but all too often, a developer simply copies and pastes the sample code from a patterns text or site and declares the result a successful application of the pattern. This is usually a recipe for failure instead of a recipe for producing a good design.
单纯死记硬背模式结构“因为权威这么说”是对亚历山大无意识设计概念的回归。如果我们这样做，就破坏了设计模式的整个目的。我们需要能够描述模式背后的原因和方法。如果不理解导致描述该模式的原因，死记硬背往往会导致误用。最好的结果是模式被破坏，与预期结果不符。最坏的情况是，它会将医源性模式注入系统 - 这种模式本应有益，但却产生了可能需要数年才能发现的恶性结果。它不仅不能提供预期的增强，还会主动产生比原问题更严重的新问题。这就是部落神话中的模式 - 不加理解的行动。
Pure rote copying of the structure of the pattern “because this authority says so” is a reversion to Alexander’s concept of unselfconscious design. We undermine the entire purpose of design patterns when we do that. We need to be able to describe the whys behind a pattern as well as the hows. Without the understanding of the reasons that led to the description of that pattern, rote application often results in misapplication. At best, the result is a broken pattern that simply does not match the intended outcome. At worst, it injects an iatrogenic pattern into the system—one that is intended and thought to be of benefit but instead produces a malignant result that may take years to uncover. It doesn’t just fail to provide the expected enhancement, it actively creates a new problem that may be worse than the original one. This is patterns as tribal mythology—action without understanding.
传统的设计模式形式（如《设计模式》 [ 21 ] 中所定义）解释了模式背后的原因（动机、适用性和后果），但读者需要自己去梳理出构成模式的底层概念。在某种程度上，这些子概念在每个模式的参与者（各个部分是什么）和协作（它们如何关联）部分中都有描述，但同样，开发人员经常将它们视为解决方案各个部分的清单，以供死记硬背实施，而不是对构成解决方案的底层概念和抽象的描述。
The traditional design pattern form, as defined in Design Patterns [21], explains the whys behind a pattern—motivations, applicability, and consequences—but it is up to the reader to tease out the underlying concepts that form a pattern. To some degree, these subconcepts are described in the Participants (what are the pieces) and Collaborations (how do they relate) sections for each patterns, but again, these are frequently treated by developers as checklists of pieces of the solution for rote implementation instead of as a description of the underlying concepts and abstractions that comprise a solution.
1.2.2. 语言相关视图
1.2.2. Language-Dependent Views
如果问开发人员模式对于其工作有多重要，答案通常取决于开发人员所使用的实现语言。这并不奇怪。不同的语言围绕它们支持的概念及其表达方式提供不同的优势。这些概念的表达方式往往是语言爱好者之间争论的开始，但在大多数情况下，忽略底层概念会导致许多无关紧要的争论。无论是像 C 系列语言那样用花括号来划分块，还是像 Python 那样用空格来划分块，都远不如首先拥有块的概念重要。
Ask a developer how important patterns are to his or her work, and frequently the answer will be based on the implementation language the developer is using. This isn’t surprising. Different languages offer different strengths centered around the concepts they support and how they express them. How those concepts happen to be expressed is more often the start of flame wars between language fans, but ignoring the underlying concepts leads to much argument over nothing of particular consequence in most cases. Whether blocks are delineated by curly braces, as in the C family, or by whitespace, as with Python, isn’t nearly as important as having the concept of blocks in the first place.
这意味着某些模式在某些语言中比在其他语言中更容易实现。事实上，有些语言可以使某些模式背后的概念非常容易实现，以至于它们被称为语言特性。Visitor模式就是一个很好的例子。1 Visitor的实现部分 [ 21，第 338 页] 写道：“ Visitor通过使用一种称为双重分派的技术来实现 [它的目标]。这是一种众所周知的技术。事实上，一些编程语言直接支持它（例如 CLOS）。”这是什么意思？这意味着向 CLOS（通用 LISP 对象系统）开发人员提及Visitor设计模式会让他们摸不着头脑。“模式？语言特性？为什么？”在 CLOS 中，Visitor基本上是内置的。你不需要一个模式来告诉你如何最好地表达这个概念——它已经作为基本特性存在于语言中了。然而，在大多数其他语言中，Visitor 提供了一种清晰的方式来表达同样的双重分派编程概念。
What this means is simply that some patterns are easier to implement in some languages than in others. In fact, some languages can make the concepts behind certain patterns so simple to implement that they’re known as language features. The Visitor pattern is a good example.1 Visitor’s Implementation section [21, pg. 338] says, “Visitor achieves [its goal] by using a technique called double-dispatch. It’s a well-known technique. In fact, some programming languages support it directly (CLOS, for example).” What does this mean? It means that mentioning the Visitor design pattern to CLOS (Common LISP Object System) developers will leave them scratching their heads. “A pattern? For a language feature? Why?” In CLOS, Visitor is essentially built in. You don’t need a pattern to tell you how to best express the concept—it’s already there in the language as a basic feature. In most other languages, however, Visitor provides a clean way of expressing the same programming concept of double dispatch.
这说明了一个重要问题。如果你向同样的 CLOS 开发人员提及双重分派而不是访问者模式，他们会知道你的意思，如何使用双重分派，以及何时不使用它。术语，尤其是共享的通用术语，非常重要。
This illustrates an important point. If you mention double dispatch instead of the Visitor pattern to the same CLOS developers, they would know what you mean, how to use double dispatch, and when not to use it. Terminology, particularly shared common terminology, matters a great deal.
所有语言和所有模式都是如此：某些语言使某些模式更容易或简单地实现，而其他模式则更难实现。然而，在这种情况下，没有一种语言可以被认为比另一种语言更好。一个普遍的误解是设计模式可以弥补编程语言的缺陷，但事实并非如此。设计模式描述了有用的概念，无论使用哪种语言来实现它们。特定概念是嵌入到语言的功能集中还是必须从头开始实现并不重要。概念仍然在实现中得到表达，这是让我们独立于软件实现来谈论软件设计的关键观察结果。设计是概念；如何在给定语言中将这些概念具体化是实现。
This is true for all languages and all patterns: some languages make certain patterns easier or trivial to implement and other patterns more difficult to realize. No language can really be considered superior to another in this case, however. One common myth is that design patterns make up for flaws in programming languages, but that isn’t the case. Design patterns describe useful concepts, regardless of the language used to implement them. Whether a specific concept is baked into the feature set of a language or must be implemented from scratch is irrelevant. The concept is still being expressed in the implementation, and that is the critical observation that lets us talk about software design independently of software implementation. Design is concepts; how those concepts are made concrete in a given language is implementation.
深入研究后，你会发现没有理由不能用纯 C 语言实现 GoF 文本中的每个模式——但这将是极其繁琐的。你必须建立最佳实践，将数据和函数绑定到有意义的语义单元中，封装数据，确保数据在首次可访问时即可使用，等等。这听起来工作量很大，但这些概念被认为非常重要，以至于它们引发了语言特性的革命，使它们更易于使用。这场革命就是面向对象编程。
When you get down to it, there’s no reason you couldn’t implement every pattern in the GoF text in plain C—but it would be extremely tedious. You’d have to build up best practices for binding data and functions into meaningful semantic units, encapsulating that data, ensuring that data is ready to use at first accessibility, and so on. This sounds like a lot of work, but these were concepts considered so important that they launched a revolution in language features to make them easier to work with. That revolution was object-oriented programming.
在面向对象语言中，这些概念被包含在称为类、可见性和构造函数的主要语言特性中。同样，我们可以参考 GoF：“如果我们假设是过程语言，我们可能会包括称为‘继承’、‘封装’和‘多态性’的设计模式。”作者认为这句话非常重要，以至于它出现在引言的1.1 节中。然而，这是一个大多数开发人员似乎都忽略的基本点，所以让我重申一下。
In object-oriented languages, those concepts are included as primary language features called classes, visibility, and constructors. Again, we can refer to the GoF: “If we assumed procedural languages, we might have included design patterns called ‘Inheritance,’ ‘Encapsulation,’ and ‘Polymorphism.’” The authors felt that this statement was important enough that it appears in Section 1.1 in the Introduction. And yet again, this is a fundamental point that seems lost on most developers, so let me restate it.
模式是与语言无关的概念；只有当您使用给定的一组语言特征和结构在特定语言中实现它们时，它们才会成形并成为具体的解决方案。
Patterns are language-independent concepts; they take form and become concrete solutions only when you implement them within a particular language with a given set of language features and constructs.
这意味着，谈论“Java 设计模式”、“C++ 设计模式”、“WebSphere 设计模式”等有点奇怪，尽管我们都这么做。这是一种略显偷懒的简写形式，代表了我们真正的意思，或者应该的意思：用 Java、C++、WebSphere 等实现的设计模式，无论语言或 API 是什么。2
This means that it is a bit strange to talk about “Java design patterns,” “C++ design patterns,” “WebSphere design patterns,” and so on, even though we all do it. It’s a mildly lazy form of shorthand for what we really mean, or should mean: design patterns as implemented in Java, C++, WebSphere, and so on, regardless of language or API.2
不幸的是，如果你像许多开发人员一样，读过大量关于设计模式的书籍，你可能被训练过，或者至少被错误地引导相信，存在一些短暂而基本的用 Java 表达的模式与用 Smalltalk 等其他语言表达的模式之间没有区别。实际上没有区别。概念是相同的；只是表达方式和程序员用特定语言实现它们的难易程度不同。
Unfortunately, if you’re like many developers who have encountered one of the multitude of books on design patterns, you may have been trained, or at least have been erroneously led to believe, that there is some ephemeral yet fundamental difference between patterns as expressed in Java and those expressed in another language such as Smalltalk. There really isn’t. The concepts are the same; only the manner in which they are expressed and the ease with which a programmer can implement them in that specific language differ.
在研究设计模式时，我们需要关注这些，这些抽象必须是理解模式的关键。除非我们努力将模式视为独立于语言的概念，否则我们只是在重新学习死记硬背的配方，并失去它们如此有用的许多东西。
We need to focus on these when investigating design patterns, and these abstractions must be the crux of understanding patterns. Unless we make the effort to look at patterns as language-independent concepts, we are merely learning rote recipes again and losing much of what makes them so useful.
1.2.3. 从神话到科学
1.2.3. From Myth to Science
前面描述的问题掩盖了设计模式的根本问题，因为如今设计模式经常被传达、使用和理解。很多时候，即使我们在代码中使用了设计模式，我们仍然不知道我们为什么要这样做。通过如此不灵活地使用设计模式，我们只是更好地记录了一组无意识的代码片段，而没有从对代码片段进行系统分析中获得理解。
The issues described previously belie an underlying problem with design patterns as they are often conveyed, used, and understood today. All too often, we still don’t know why we do what we do, even when we use design patterns in our code. By using design patterns so inflexibly, we’ve simply better documented a body of unselfconscious snippets without the comprehension that comes from a methodical analysis of the snippets.
我们拥有一门艺术。我们需要的是一门科学。毕竟，我们肆意使用计算机科学和软件工程这两个术语。将模式视为示例代码会忽略设计模式的要点。设计模式使我们作为一个行业能够试验这些概念并分享、讨论和改进我们的发现。
We have an art. What we need is a science. After all, we throw around the terms computer science and software engineering with abandon. Treating patterns as sample code misses the point of design patterns. Design patterns enable us as an industry to experiment with those concepts and share, discuss, and refine our findings.
作为死记硬背的菜谱的模式是部落神话。
Patterns as rote recipes are tribal mythology.
模式作为概念是科学的基础。
Patterns as concepts are the foundations of a science.
基本设计模式是该科学的基石。
Elemental Design Patterns are the building blocks of that science.
2.元素设计模式
2. Elemental Design Patterns
基本设计模式 (EDP) 本质上是面向对象编程基本概念的目录。它们有两个特点：首先，它们是以设计模式文献的风格编写的。每一个都被视为一个独立的概念，有一个特定的名称，可以通过这个名称进行讨论和思考，直到理解它。本书介绍了每个 EDP 以及它适用的问题，并讨论了何时应该和不应该应用该模式。您将找到示例实现、对使用这些模式的可能后果的评论，以及您也应该查看的相关模式。每个 EDP 的这种以人为本的定义为您提供了一个通用术语，您可以使用它清楚而准确地与其他学生或开发人员讨论这个概念。
The Elemental Design Patterns, or EDPs, are, at their core, a catalog of the basic concepts of object-oriented programming. Two traits make them unique: First, they are written in the style of the design patterns literature. Each is treated as a standalone concept with a specific name by which it can be discussed and mulled over until it is understood. This book presents each EDP along with a problem that it is suited for and discusses when the pattern should and should not be applied. You will find example implementations, comments on possible consequences of using the patterns, and related patterns that you should look at as well. This human-oriented definition of each EDP provides you with a common term you can use to clearly and precisely converse with other students or developers about the concept.
其次，设计模式是对通过检查现有软件系统发现的常见问题的解决方案的描述。EDP 也是常见问题的解决方案，一旦你知道要寻找什么，你就会发现它们无处不在。事实上，它们是如此常见，以至于到目前为止，它们还没有被认为值得以任何全面的方式写下来。EDP 最初被认为是软件中相互关联的设计概念的综合体，而不是软件中的设计概念，而是软件的正式描述。EDP 源自数学基础面向对象语言，但请放心，除了安全地藏在后面的附录外，本书不含数学。（但我希望您好奇并阅读附录——它相当有趣。）EDP 背后的坚实而简单的思想提供了一个强大的框架，可以在此框架上重新审视这些基本概念，赋予它们新的生命，并将它们重新用于新的应用。这个基础还使您能够将它们用作微小的构建块，几乎像原子一样，以坚实而直接的方式描述其他设计模式。而且，就像原子一样，您可以用它们构建世界。
Second, design patterns are descriptions of solutions to common problems that were found through the examination of existing software systems. EDPs are also solutions to common problems, and once you know what to look for, you will find them everywhere. They are in fact so common that, until now, they have not been considered worth writing up in any comprehensive way. EDPs were originally identified as a comprehensive body of interrelated design concepts not in software but in a formal description of software. EDPs arise from the mathematical foundations of object-oriented languages, but rest assured, with the exception of an appendix safely tucked in the back, this text is mathematics free. (I do hope you get curious and read the appendix, though—it’s rather interesting.) The solid yet simple ideas underlying the EDPs provide a strong framework on which to revisit these basic concepts, give them new life, and repurpose them for new applications. This foundation also gives you a compelling ability to use them as tiny building blocks, almost like atoms, to describe other design patterns in solid and direct ways. And, like atoms, you can build worlds with them.
本书将向您介绍 EDP 的一些背景知识，包括其起源，以及它们如何融入编程中的设计模式大背景。它还将让您对面向对象理论有初步的了解，让您了解所有模式如何相互关联，但我保证，本书不会提供任何等式。我们将讨论面向对象编程基础的一些基本思想，其中许多思想您可能已经很熟悉。最后，您将了解该基本理论如何产生一系列可用于日常编程的概念。后面的章节将解释如何使用 EDP 来改善您的设计。
This book gives you some background on EDPs, where they came from originally, and how they fit into the larger context of design patterns in programming. It also gives you just enough of a taste of object-oriented theory to understand how the patterns all relate to one another, but no equations, I promise. We discuss some of the ideas underlying the basics of object-oriented programming, much of which you may already be familiar with. Finally, you learn how that basic theory gives rise to a veritable stable of concepts that you can use in your everyday programming. Later chapters explain how you can use EDPs to make your designs better.
2.1. 背景
2.1. Background
EDP 的识别起源于北卡罗来纳大学教堂山分校的模式查询和识别系统 (SPQR) 项目 [ 35 ]。SPQR 是一个在现有源代码中识别已知设计模式实例的研究项目。它可以查找独立于原始源代码语言的设计模式，并可以轻松地从单个模式定义中找到同一设计模式的各种实现。如果你对 SPQR 的内部工作原理感兴趣，可以查阅更多信息 [ 35、37、38 ] ，但基本的了解已经为我们讨论什么是 EDP 以及如何定义 EDP 提供了足够的背景信息。
The identification of EDPs had its beginnings in the System for Pattern Query and Recognition (SPQR) project at the University of North Carolina at Chapel Hill [35]. SPQR is a research project for identifying instances of known design patterns within an existing body of source code. It can find design patterns independent of the original source code language and can easily find various implementations of the same design pattern from a single pattern definition. If you’re interested in the inner workings of SPQR, further information is available [35, 37, 38], but a basic understanding provides enough context for our discussion of what EDPs are and how they are defined.
我在担任专业软件工程师期间萌生了创建 SPQR 的想法。当时我所在的团队负责实时商用和军用飞行模拟系统中使用的三个库之一。在我们与应用程序团队的一次联合全库开发会议上，他们非常感谢我们刚刚添加的一项功能，并表示这正是他们所需要的。我们说他们非常受欢迎，这就是我们来这里的目的，等等。
I was inspired to create SPQR while working as a professional software engineer. I was on a team responsible for one of three libraries used in a real-time commercial and military flight simulation system. In one of our joint all-library development meetings with the application team, they thanked us profusely for a feature we had just added, stating that it was exactly what they needed. We said they were quite welcome, that’s what we were there for, and so on.
会议结束后，三个图书馆团队互相看了看，询问其他人是否知道应用程序团队在说什么。我们中没有人设计过他们描述的功能，我们甚至不确定它是怎么来的不知道，或者说不可能。我们决定彻底查明真相，认为这肯定不会花太多时间。
After the meeting, the three library teams looked at each other and asked if the others knew what the application team had been talking about. None of us had designed the feature they were describing, we weren’t even sure how it had come about, or that it was possible. We decided to get to the bottom of it, thinking that certainly it couldn’t take much time.
经过近 200 个开发小时，我们终于找到了答案。我们有一个装饰模式实例，它是四人帮 (GoF) 模式之一，嵌入并隐藏在系统中。独特之处在于它不存在于任何一个库中。它的各个部分分散在三个库中，应用程序团队偶然发现了它们作为一个整体的集成。我们惊呆了。随后发生了一场相当激烈的争论：这真的是一种设计模式吗？毕竟，它不是设计出来的，只是逐渐发展起来的。
Nearly 200 developer hours later, we had our answer. We had an instance of a Decorator pattern, one of the Gang of Four (GoF) patterns, embedded and hidden within the system. What was unique was that it didn’t live in any one library. Pieces of it were scattered across the three libraries, and the application team had stumbled onto their integration as a whole. We were stunned. A rather heated debate ensued: Was this really a design pattern? After all, it hadn’t been designed, it had just sort of grown.
嗯，大多数软件都是这样。它有机地发展，而且往往以意想不到的方式发展。在管理软件项目时，我们只能管理我们所知道的东西，知道有这个有用的设计元素为我们提供了一个机会来增强、简化和记录它的存在，以便开发人员有效地使用。找到它——这就是问题所在。记住，我们编写了有问题的软件，我们是专家，但我们仍然花了大量的时间来推断发生了什么。流程自动化是显而易见的答案。正如我在第 1 章中提到的，我们真正想要的是自文档化代码，或者至少是一个提取该文档的系统。SPQR 就是这样一个系统。
Well, most software does just that. It grows organically, and often in unexpected ways. In managing software projects, we can only manage what we know, and knowing that there was this useful design element provided us an opportunity to enhance, streamline, and document its existence for developers to use effectively. Finding it—that was the problem. Remember, we had written the software in question, we were the experts, and it still took us an inordinate amount of time to deduce what was going on. Automation of the process was the obvious answer. As I mentioned in Chapter 1, what we really want is self-documenting code, or at least a system to extract that documentation. SPQR is such a system.
创建 SPQR 需要解决一个基本问题：我必须教会计算机如何识别设计模式。模式不是死记硬背的处方；它们是我们人类称之为概念的柔软而无定形的东西。大多数试图在源代码中查找模式实例的研究和行业系统都是通过将模式视为构造（看起来像实现的刚性形式）而不是抽象概念来实现的。这是一种合理的方法，它像许多开发人员一样对待模式：“这个类有那个字段，那个子类通过名为X 的方法访问它”，等等。问题是，这种方法再次将设计模式简化为特定的实现。可能实现的每种变体都需要新的定义。这又是剪切粘贴方法 [ 25，30 ] 。我并不是说这项工作质量低下——远非如此。这些工具是该方法的杰出示例，值得您研究。它们做得很好，如果您碰巧满足它们的要求，它们会增加价值。
Creating SPQR necessitated solving a fundamental problem: I had to teach the computer how to identify design patterns. Patterns are not rote recipes; they are soft and amorphous things we humans call concepts. Most research and industry systems that attempt to find instances of patterns in source code do so by looking at patterns as constructs—as rigid forms that look like implementations—rather than as abstract concepts. This is a reasonable approach, and it treats patterns as many developers do: “This class has that field, and that subclass accesses it through a method named X,” and so on. The thing is, this approach once more reduces the design patterns to specific implementations. Every variation of a possible implementation requires a new definition. It’s the cut-and-paste methodology again [25, 30]. I don’t mean that this work is of low quality—far from it. These tools are stellar examples of the approach and are worth your investigation. They do their job well, and they add value if you happen to meet their requirements.
这种方法的问题在于，将查询模式定义为构造非常脆弱。如果你定义的特定方法不是名为X而是名为Y会怎么样？如果子类不是直接子类，但在继承层次结构中两者之间有另一个类会怎么样？或者，如果子类根本不是真正的继承子类，而是使用超类作为委托对象会怎么样？面向构造的方法不会发现这些情况中的任何一个设计模式，它们将需要新的模式定义。而且这是假设您仍在使用一种实现语言。请记住，模式与实现语言无关——问题的范围要大得多。
The problem with this approach is that defining a pattern for queries as a construct is highly fragile. What if the particular method you define wasn’t named X, but Y? What if the subclass wasn’t a direct subclass, but there was another class between the two in the inheritance hierarchy? Or what if the subclass wasn’t a true inherited subclass at all but used the superclass as a delegate object instead? None of these conditions would be found by a construct-oriented approach to design patterns, and they would require new pattern definitions. And that’s assuming you’re still working within one implementation language. Remember, patterns are implementation-language independent—the scope of the issue is much larger.
显然，SPQR 需要一种不同的策略。深入研究设计模式的背景和历史后，我找到了必要的线索。我研究 GoF 材料越多，回溯产生它的灵感，我就越关注亚历山大的原创作品。当我阅读他的第一篇关于设计模式的论文《形式综合笔记》[ 4 ]时，我意识到一些重要的东西已经被遗漏了：模式是概念，而不是构造。
It seemed obvious that a different tactic was needed with SPQR. Diving into the background and history of design patterns yielded the necessary clue. The more I studied the GoF material and worked back through the inspirations that spawned it, the more I focussed on Alexander’s original work. As I read his first and, at least within the software engineering community, least-cited treatise on design patterns, Notes on the Synthesis of Form [4], I realized something important had been lost: patterns are concepts, not constructs.
这个简单的事实从根本上改变了我试图教计算机识别的东西。教计算机系统寻找用刚性结构固定在一起的语言结构永远不会给我们带来我们想要的灵活性和准确性。相反，我们应该教系统寻找编程概念，因为那才是真正的模式。
This simple truth fundamentally alters what I was trying to teach the computer to recognize. Teaching a computer system to look for language constructs bolted together in rigid structures will never give us the flexibility we desire with the accuracy we need. Instead, we should be teaching a system to look for programming concepts because that’s what patterns really are.
不幸的是，对于这项特定任务，现有的设计模式文献主要侧重于描述高级抽象概念，这也是它们如此强大的原因之一——它们将设计的讨论提升到更高的抽象层次。程序员通常能够很好地掌握较低级别的抽象，因此将它们建立为模式似乎从未必要。设计模式社区正确地专注于记录那些尚未根深蒂固和被充分理解的课程。软件社区描述的这些设计模式往往处于频谱的较大端。另一方面，计算机的知识从根本上来说只取决于我们能教给它们的基础。对于 SPQR，我们必须从头开始建立一系列概念，以便它能够帮助我们处理编程概念。我们必须填补实现的高度僵化部分与它们所代表的有时流动的概念之间的空白。
Unfortunately for this particular task, the established design patterns literature is primarily geared toward describing high-level, abstract concepts, which is part of what makes them so powerful—they lift the discussion of a design to a higher level of abstraction. Programmers generally have a good working handle on lower-level abstractions, so establishing them as patterns has never seemed necessary. The design patterns community is rightly focused on documenting those lessons that are not already ingrained and well understood. These design patterns described by the software community tend to be at the larger end of the spectrum. Computers, on the other hand, are fundamentally only as knowledgable as the foundation we can teach them. For SPQR, we had to establish a chain of concepts from the ground up so it could help us deal with the concepts of programming. We had to fill in the gap between the highly rigid pieces of an implementation and the sometimes fluid concepts that they represent.
通过这样做，我们可以看出 EDP 就是这样的设计模式，它构成了描述软件设计的基本基础，不仅适用于自动分析，也适用于人类。本书是为人类而非机器使用而编写的 EDP 的部分目录。它旨在供开发人员、学生和设计人员使用，以帮助填补我们在讨论软件设计时在理解和语言上的空白。正如设计模式所预期的那样，每个 EDP 都有一个非正式的编写或模式规范（使用一个公认的术语），并且也起源于数学构造。后一个特性使 EDP 目录独一无二。每个 EDP 都是编程的抽象，让我们可以谈论甚至非常小的软件设计问题也与其具体的实施方式无关。
In doing so, it became apparent that the EDPs were just that—design patterns that form a fundamental basis from which to describe software design not just for automated analysis but for humans as well. This book is a partial catalog of the EDPs written for human, not machine, consumption. It is intended to be used by developers, students, and designers to help fill in the gaps in our understanding and our language when discussing software design. As expected with design patterns, each EDP has an informal write-up, or pattern specification, to use an accepted term, and also has an origin as a mathematical construct. This latter trait makes the EDP catalog unique. Each EDP is an abstraction of programming that lets us talk about even very small issues of software design independent of precisely how they are implemented.
这给我们带来了两种描述 EDP 的方式：它们的理论基础和实用主义起源。下一节将两者交织在一起，同时将形式主义保持在最低限度。您不需要在数学层面上准确理解形式基础才能继续阅读本书，但如果您希望朝这个方向前进，您可能会发现它很有启发性。如果您对底层机制感到疑惑，并希望更深入地了解设计模式和编程语言理论如何相交，您可以从附录开始，其中描述了构成这项工作基础的 ρ 演算。
This brings us to two ways of describing EDPs: their basis in theory and their genesis in pragmatism. The following section intertwines the two while keeping the formalisms to a minimum. You don’t need to precisely understand the formal foundation at a mathematical level to continue in this book, but you may find it illuminating if you wish to move in that direction. If you find yourself wondering about the underlying mechanics and want a stronger understanding of how design patterns and programming language theory intersect, you can start in the appendix, which describes the ρ-calculus that forms the foundation of this body of work.
2.2. 地点、原因和方式
2.2. The Where, the Why, the How
教会计算机寻找大型抽象概念和设计模式（它们存在于常见文献中）是一项艰巨的任务。面对计算机科学中一项庞大而棘手的任务时，我们该怎么做？我们将其分解。
Teaching a computer to look for the large-scale abstractions and concepts called design patterns, as they exist in the common literature, is a difficult task. What do we do when faced with a large and gnarly task in computer science? We break it down.
解构设计模式文献并非易事。多年来，人们进行过几次尝试 [ 12、17、32、40、41、43 ] ，但这些尝试都只是部分解构，大多数开发人员和领域外的研究人员都认为这些解构是奇怪或令人好奇的。毕竟，这些较小的解构部分和概念是显而易见的，对吧？它们是基本概念，是我们每天都要处理的基本事物，所以何必费心描述我们已经知道的东西呢？
Deconstructing the design patterns literature is not a simple thing. A few attempts have been made over the years [12, 17, 32, 40, 41, 43], but they have been partial deconstructions, seen as oddities or curiosities by most developers and researchers outside the immediate field. After all, these smaller deconstructed pieces and concepts are obvious, right? They are basic concepts, basic things we deal with every day, so why bother describing what we already feel we know?
那么，如果它们如此明显，为什么它们还没有被记录下来呢？
Well, if they’re so obvious, how come they’re not already documented?
软件设计的基本概念与建造住宅的“正确”方式一样“显而易见”：它完全取决于您的环境、您的经验以及您所学的内容。接受过 ML 函数式编程培训的人与主要背景是 C 的人对递归的理解截然不同。这两位开发人员对递归、递归的用途以及应如何应用递归有不同的假设，即使它解决的基本问题相同。鉴于第 1 章中对住房风格的讨论，这个想法应该开始听起来很熟悉。
The basic concepts of software design are as “obvious” as the “correct” way to build a domicile: it depends entirely on your context, your experience, and what you were taught. Someone who was trained in functional programming in ML understands recursion very differently from someone whose primary background is C. The two developers will have different assumptions about recursion, where it is useful, and how it should be applied, even though the basic underlying problem that it is solving is the same. Given the discussion of housing styles in Chapter 1, this idea should start to sound familiar.
我在本章前面说过，编程的底层概念并不是设计模式社区的主要目标，因为文献旨在记录既不根深蒂固又不为人所熟知的概念。简单的概念被视为根深蒂固，因此超出了这一范围，但这并不意味着它们实际上被很好地理解了。
I said earlier in this chapter that the lower-level concepts of programming have not been a primary target of the design patterns community because the literature is aimed at documenting concepts that are neither ingrained nor well understood. Simple concepts are seen as ingrained and therefore outside of that scope, but that doesn’t mean they’re actually well understood.
正如 Alexander 所定义的那样，我们在编程中使用的低级概念是无意识的。我们在课堂上和实践中或多或少地死记硬背这些概念，我们应用它们是因为“这就是它的工作方式”，而不需要有意识地做出决定。我们大多数人从未研究过这些概念背后的原理，但它们确实存在，并且在数学层面上很容易理解。
The low-level concepts we use in programming are unselfconscious, as Alexander defined it. We learn them more or less by rote in class and in the field, and we apply them because “that’s the way it works,” without having to make conscious decisions to do so. Most of us never study the principles underlying those concepts, but they exist and are well understood at a mathematical level.
这些概念还没有在人类层面上得到适当的展现，而 EDP 就是这样的概念。简而言之，EDP 是编程和软件设计的基本核心概念，这些概念大多尚未得到描述。到目前为止，那些已被描述的概念概念并没有以有意义且有理有据的方式相互关联。每个 EDP 都是对一个核心概念的自我描述。EDP 目录作为一个整体将 EDP 相互关联，形成一个概念框架，学生或开发人员可以使用它来理解其他模式。它提供了用于描述更高级别抽象的分类法和词汇表，使语言和抽象同质化，以便当任何两个开发人员引用例如 EDP扩展方法时，他们都对共同的定义有精确的理解。
Those concepts just haven’t been properly exposed at a human level, and that’s what EDPs are. In a nutshell, EDPs are the underlying core concepts of programming and software design that have remained mostly undescribed. Those that have been described have not been related to one another in a meaningful and well-founded manner until now. Each EDP is a selfconscious description of a core concept. The EDP catalog as a whole relates the EDPs to one another to form a conceptual framework that the student or developer can use to understand other patterns. It provides a taxonomy and lexicon for describing higher-level abstractions, homogenizing the language and abstractions such that when any two developers reference, for example, the EDP Extend Method, they both have a precise understanding of a common definition.
EDP 提供了一种语言，可用于从最基础的层面推理、描述和讨论软件。主流设计模式在为经验丰富的专业人士提供这一知识库方面做得非常出色，但到目前为止，还没有人以全面的方式将同一工具瞄准新手或学生。
EDPs provide a language with which to reason about, describe, and discuss software, from the most fundamental levels on up. Mainstream design patterns have done an amazing job of providing this knowledge base for the seasoned pro, but until now, no one has aimed that same tool at the novice or student in a comprehensive way.
那么，我们能从设计模式文献中解构出这些概念到什么程度呢？作为例子，我们来看看Decorator。
So how far can we deconstruct these concepts from the design patterns literature? As an example, let’s look at Decorator.
2.2.1.装饰器的分解
2.2.1. Decomposition of Decorator
在继续之前，我们先简单了解一下命名约定：本文中的模式名称采用斜体大写，例如Extend Method。示例中的类型和类采用大写，并使用 monotype 字体。字段和方法采用大写字母或“驼峰式”格式，并使用 monotype 字体，例如thisIsAMethod()。代码示例使用相同的系统。
A quick note on naming conventions before we go any further: pattern names in this text are italicized and capitalized—for example, Extend Method. Types and classes in examples are capitalized and set in monotype font. Fields and methods are set in intercaps, or “camelcase” and in monotype font, as in thisIsAMethod(). Code examples use the same system.
图 2.1显示了GoF 的设计模式[ 21 ] 中提供的装饰器统一建模语言 (UML) 的典型示例。现在，您不需要了解装饰器是什么；查看 UML 图就足够了。一般来说，装饰器是一种流行且常用的设计模式，它提供了一种在运行时动态扩展行为的机制。您可以将其视为内部插件或扩展系统，就像在 Web 浏览器中找到的那样。
The canonical example Unified Modeling Language (UML) for Decorator, as provided in the GoF’s Design Patterns [21], is shown in Figure 2.1. For now, you don’t need to understand what Decorator is; a visual inspection of the UML diagrams should suffice. In general, Decorator is a popular and commonly used design pattern that provides a mechanism for behavior to be extended dynamically at runtime. You can think of it as an internal plug-in or extension system, like you would find in a web browser.
图 2.1.装饰器的常用示例 UML。
Figure 2.1. Decorator’s usual example UML.
假设您想分解Decorator以更好地理解它。毕竟，它是一种相当高级的抽象。如果您可以单独吸收较小的部分，那么您几乎肯定可以更轻松地理解该模式。更好的是，您应该能够使用这些较小的部分来帮助您理解您可能遇到的其他设计模式，只要它们以类似的术语描述即可。
Suppose you want to decompose Decorator to better understand it. It is, after all, a rather high-level abstraction. If you can absorb smaller pieces individually, you can almost certainly have an easier time understanding the pattern. Better still, you should be able to use those smaller pieces to help you understand other design patterns you may encounter, as long as they’re described in similar terms.
这与试图了解赛车等复杂机械的工作原理没有什么不同。您可以购买一辆赛车，将其放在车库中，取出工具，拆开整辆车，也可以单独了解各个部件。例如，您可以研究内燃机或液压制动系统。这些系统可以进一步分解成各个部分，您可以逐一研究，从而使更大的研究任务更容易。
It’s not unlike trying to understand how a complex piece of machinery such as a race car works. You can buy one, put it in your garage, break out the tools, and disassemble the whole car, or you can learn about the pieces individually. You can study the internal combustion engine, for instance, or the hydraulic braking system. Those systems can be further broken down into parts that you can investigate one by one to make the larger study task easier.
最后，如果你知道汽油发动机的工作原理，你不仅能够理解赛车的这一部分，将其作为一个整体，一个封装的抽象概念，还可以将这些知识应用于其他车辆，甚至是割草机、发电机和任何其他汽油发动机驱动的机器。分解设计模式的目的也类似：识别出你可以根据自身优点进行处理的较小部分，以更易于理解的方式进行彻底学习，并应用于新情况。
In the end, if you know how a gasoline engine works, you not only have the facility to comprehend that portion of the race car as a unit, as an encapsulated abstraction, but you can apply that knowledge to other vehicles, or even lawnmowers, generators, and any other gasoline-engine-driven machine. Decomposing design patterns serves a similar purpose: identifying the smaller pieces that you can treat on their own merits, learn thoroughly in more digestible chunks, and apply in novel situations.
在对现有的设计模式文献进行详尽搜索之后（请记住，有数千个），您可能会注意到Decorator与另外两种模式有相似之处。
After an exhaustive search of the existing literature on design patterns (and remember, there are thousands), you might notice that Decorator shares similarities with two other patterns.
一种是对象化器，由 Walter Zimmer 于 1995 年首次提出 [ 43 ]，如图 2.2所示。对象化器描述了一种用单个对象接口来表示多个以对客户端不透明的方式调用具体实现。当客户端通过Objectifier接口请求调用某个方法时，它并不知道要执行的是两个（或更多）具体方法体中的哪一个。
One is Objectifier, first described by Walter Zimmer in 1995 [43], shown in Figure 2.2. Objectifier describes a way for a single object interface to represent multiple concrete implementations in a way that is opaque to a client. When a client requests a method to be invoked through the Objectifier interface, it does not know which of the two (or more) concrete method bodies is executed.
图 2.2.对象化器作为 UML。
Figure 2.2. Objectifier as UML.
另一种是对象递归，由 Bobby Woolf 于 1998 年提出[ 41 ]，如图 2.3所示。对象递归将两个具有相关类型的对象链接在一起。
The other is Object Recursion, outlined by Bobby Woolf in 1998 [41], shown in Figure 2.3. Object Recursion chains together two objects with related types.
图 2.3.对象递归作为 UML。
Figure 2.3. Object Recursion as UML.
查看图 2.2和2.3中的 UML 图，您会发现Objectifier 和 Object Recursion非常像Decorator模式的某些部分。尽管它们并不完全相同，但它们的 UML 结构中的某些特征非常相似，因此我们可以用这些其他模式来描述Decorator的某些部分 — — 我们可以说Decorator是由这些模式组成的。
Looking at the UML diagrams in Figures 2.2 and 2.3, you can see that Objectifier and Object Recursion look quite a bit like pieces of the Decorator pattern. Although they are not exactly the same, certain features in their structure of the UML are similar enough that we can describe parts of Decorator in terms of these other patterns—we can say that Decorator is composed of these patterns.
仔细观察就会发现，Objectifier可以看作是Object Recursion的一部分。Object Recursion使用Objectifier作为其形式的骨干，但它在具体实现和接口之间添加了链接，并且确实对于相同的方法也是如此。换句话说，当Initiator类调用时，按照ObjectifierhandleRequest()的意图，或类可能会处理它。如果处理它，则请求完成。如果处理它，则通过接口对另一个进行额外调用，然后重新开始该过程。此链一直持续到处理程序，此时链不出所料地终止。TerminatorRecursorTerminatorRecursorhandleRequest()Terminator
A closer look shows that Objectifier can be considered a part of Object Recursion. Object Recursion uses Objectifier as the backbone of its form, but it adds a link between one of the concrete implementations and the interface, and it does so for the same method. In other words, when the Initiator class calls handleRequest(), by the intent of Objectifier, either the Terminator or Recursor class might handle it. If Terminator handles it, the request is complete. If Recursor handles it, an additional call is made through the interface for another handleRequest(), and the process starts over again. This chain continues until Terminator is the handler, at which point the chain, unsurprisingly, terminates.
鉴于这一系列操作，我们可以说：“装饰器使用对象递归来遍历一个对象链，大概至少有两个对象，并对每个对象执行相同的方法调用。”但这似乎相当模糊，而且也遗漏了装饰器中的许多细节。我们应该有更好的分解方法，但至少我们知道可以用更小的模式来描述模式。
Given this series of actions, we can say, “A Decorator uses Object Recursion to traverse a chain of objects, presumably at least two in length, and perform the same method call on each.” That seems rather vague, though, and it also misses a lot of the detail in Decorator. We should have a better decomposition to work with, but at least we know that patterns can be described in terms of smaller patterns.
2.2.2. 进入兔子洞
2.2.2. Down the Rabbit Hole
我们已经确定可以用一些较小的块来构建装饰器，但这些块是否已经小到可以再小的程度了呢？我们可以制作的最小解构模式是什么，但仍然是一个模式？
We’ve established that we can build Decorator from some smaller pieces, but are these pieces as small as they can be? What is the smallest deconstructed pattern we can make that is still a pattern?
要回答这个问题，我们需要再问一遍，什么是模式？我们知道它是一个概念，我们知道它是设计的一个元素，我们知道它有某些关键组成部分。让我们先把概念是什么的问题放在一边，讨论其他两个部分。设计是整体各部分相互作用和相互关联的方式，正如 GoF 文本中普遍描述的以公认规范格式编写的设计模式大纲所示。GoF 的大纲表明，参与者和协作是设计的核心部分，它们强调了设计的各个部分以及它们如何相互作用和相互关联。
To answer that question, we need to ask again, What is a pattern? We know it is a concept, we know it is an element of design, and we know has certain critical components. Let’s set aside the question of what a concept is and address the other two portions. Design is the manner in which parts of a whole interact with and relate to one another, as illustrated in the outline of a design pattern written in the accepted canonical format popularly described in the GoF text. GoF’s outline shows that participants and collaborations are core pieces of a design, and that they highlight both the parts of the design and how they interact with and relate to each other.
考虑一下设计模式的一个常见定义： “特定上下文中常见问题的常见解决方案”。规范格式的设计模式规范的大部分内容可以轻松地归入以下三个类别：解决方案（或实现）、问题（或描述）和上下文（或环境），如表 2.1所示。结构、实现和示例代码无疑是解决方案。意图、动机和已知用途描述了问题空间。上下文，即发生问题的环境，指导适用性，框架对可能后果的讨论，并且通常指示当前模式是否不完全正确以及哪种相关模式是更好的选择。
Consider a common definition of design pattern: “A common solution to a common problem within a particular context.” Most of the pieces of a canonical format design pattern specification can easily be placed into those three categories: solution (or implementation), problem (or description), and context (or environment), as in Table 2.1. The structure, implementation, and sample code are unmistakably the solution. The intent, motivation, and known uses describe the problem space. The context, the environment in which the problem occurs, guides applicability, frames the discussion of possible consequences, and often indicates if the current pattern isn’t exactly right and which of the related patterns is a better choice.
Table 2.1. Pattern pieces sorted into three categories of a pattern definition
对这些元素进行分类后，我们剩下两个部分，它们很难归入上述三个类别之一：参与者和协作。它们显然是解决方案的一部分，因为它们共同构成了解决方案的大部分。但它们也是解决方案的一部分问题描述，因为它们几乎总是直接反映问题的表述方式。最后，它们是部分上下文，因为它们是根据所处环境的需求而创建的。它们描述了设计模式三个分支交叉处的部件及其关系。
After sorting these elements we have two parts left that don’t easily fit into one of those three categories: participants and collaborations. They’re obviously part solution in that together they form much of the solution. But they’re also part problem description because they almost always directly reflect how the problem is phrased. Finally, they’re part context in that they are created in response to the needs of the environment in which they are placed. They describe the parts and their relationships that lie at the intersection of the three arms of a design pattern.
关系是设计的核心。汽车的设计不仅仅是一堆零件：它还是它们如何组合在一起的蓝图。房屋不仅仅是一堆木材、一盒钉子和一些铜管；它具有抵抗熵的形式或计划，我们可以说，它使结构的能量水平高于一堆瓦砾。零件清单确保您拥有开始建造所需的一切，但设计中的关系会告诉您如何使它们协同工作。
Relationships form the core of design. The design of a car is more than a pile of pieces: it is also a blueprint for how they fit together. A house is more than a stack of lumber, a case of nails, and some copper piping; it has a form, or plan, that fights entropy and, we could say, keeps the structure at a higher energy level than just a rubble pile. The parts list ensures that you have everything you need to begin building, but the relationships in the design tell you how to make them work in concert.
让我们稍微反转一下这个问题：我们可以定义的最小关系是什么？嗯，这个问题很简单。两个事物之间的单一关系是最简单的。现在我们可以将这一见解应用于设计模式的解构。
Let’s flip the question around a bit: What is the smallest relationship we can define? Well, that one is easy. A single relationship between two things is simplest. Now we can apply this insight to the deconstruction of design patterns.
这给了我们一个明确的目标。对于每个后续较小的设计模式，我们都可以用批判的眼光来审视它，并问自己：“这是否体现了不止一种关系？”如果没有，那么我们肯定已经达到了目标。如果是这样，我们仍然有可能做到了。
This gives us a clear goal. With each subsequent smaller design pattern, we can look at it with a critical eye and ask, “Does this embody more than one relationship?” If not, then we’ve definitely reached our goal. If so, it’s still possible that we’ve done so.
如果涉及多个关系，为什么我们“可能”已经完成了？好吧，并非所有关系都是平等的。有些关系在讨论设计时非常重要，有些关系则用于提供上下文。主要的上下文关系是范围界定，它以多种形式出现。
Why would we already “possibly” be done if there is more than one relationship involved? Well, not all relationships are created equal. Some are quite important when discussing design, and some are there to provide a context. The primary contextual relationship is that of scoping, and it appears in many forms.
无论何时声明变量、定义方法或描述类，新项“存在”的范围都称为其作用域。作用域是使该元素与系统中所有其他元素不同的方式。如果您有两个类，每个类都名为Menu，但一个类在名为的包中定义GraphicalUIElements，另一个类在名为的包中定义在名为 的包中定义RestaurantNecessities，您可以肯定它们不是同一个东西。它们的作用域表明它们是不同的类，绝不会混淆。此原则适用于任何时候，有一个封闭的东西，它有名称，并且您在其中定义新的东西。
Whenever you declare a variable, define a method, or describe a class, whatever that new item “lives” in is called its scope. The scope is how that element is made unique from all others in the system. If you have two classes, each named Menu, but one is defined in a package named GraphicalUIElements and the other is defined in a package named RestaurantNecessities, you can be pretty sure that they’re not the same thing. Their scope indicates that they are distinct classes, never to be confused. This principle applies any time there is an enclosing something that has a name and within which you define something new.
类是它们定义的方法和字段的作用域。命名空间和包是其中任何内容的作用域。方法和函数是其中声明的局部变量的作用域。我们一次又一次地看到相同的机制在许多不同的语言特性中发挥作用。要访问特定元素，我们通过从上到下提供作用域来精确地说明我们想要哪个元素。不幸的是，事情并不总是这么简单。有时这些作用域是隐式的，我们可以省略它们，例如在引用方法中的局部变量或类中的另一个成员时。此外，尽管它们的行为相同，但这些作用域可以具有不同的访问语法。例如，在 C++ 中，要访问命名空间Menu中的类GraphicalUIElements，我们会使用双冒号将其声明为GraphicalUIElements::Menu。但是，给定该类的一个实例，我们会使用点运算符将其访问为菜单项列表aMenu.theItems。这两种技术都指定了我们希望从作用域元素中选择哪个元素，但它们以不同的方式进行。本文定义了一种处理所有范围的单一方法，无论它是如何针对特定语言中的特定语言特性实现的。
Classes are scopes for the methods and fields they define. Namespaces and packages are scopes for anything inside of them. Methods and functions are scopes for the local variables declared within them. Over and over again we see the same mechanism at work in many different language features. To access a specific element, we state precisely which one we want by providing the scopes from the top down. Unfortunately, it isn’t always this simple. Sometimes these scopes are implicit, and we can leave them off, such as when referring to a local variable within a method or to another member within a class. Furthermore, despite their common behavior, these scopes can have different syntaxes through which they are accessed. For instance, in C++, to access the Menu class within the GraphicalUIElements namespace, we would state it as GraphicalUIElements::Menu, using a double colon. Given an instance of that class, however, we would access its list of menu items as aMenu.theItems, using a dot operator. Both of these techniques specify which element we wish to select from a scoping element, but they do so in different ways. This text defines a single way in which we can treat all scoping, regardless of how it is implemented for specific language features within specific languages.
考虑A具有方法的类f，以及B具有方法的类，如图2.4g中的 UML和清单 2.1中的代码所示。具有类型为的字段。在主体中，实例化类型的实例，然后调用。方法的范围由对象决定。在面向对象语言中，函数和方法总是包含在某个范围中，即使范围是隐式的。AbBmain()aAa.f()f()a
Consider class A that has a method f, and class B that has a method g, as shown in UML in Figure 2.4 and in code in Listing 2.1. A has a field b of type B. In the main() body, an instance a of type A is instantiated, and then a.f() is invoked. The method f() is scoped by the object a. In object-oriented languages, functions and methods are always enclosed by some scope, even if the scope is implicit.
图 2.4. 一个简单的方法调用作为 UML。
Figure 2.4. A simple method call as UML.
即使在 C++ 中，全局函数和字段也可以看作是驻留在代表全局命名空间的不可见隐式对象内，这就是运行时有效处理它们的方式。请查看 2011 C++ ISO 标准工作草案 [ 6 ] 的第 3.3.6 条 [basic.scope.namespace] 第 3 段进行确认。此外，现在，传统观点不鼓励在 C++ 中使用文件作用域元素（即声明为全局和静态的元素），而是倾向于使用未命名的命名空间，它们可以更明确地执行相同的功能。换句话说，每个翻译单元都有自己的对象来进行作用域，而作用域没有名称这一事实意味着它不能在该翻译单元之外使用。同样，对象被用来限定作用域和包装以前被视为自由漫游的元素。
Even in C++ global functions and fields can be considered as though they reside inside an invisible and implicit object representing a global namespace, which is how the runtime effectively treats them. Check Clause 3.3.6 [basic.scope.namespace], paragraph 3, of the 2011 C++ ISO Standard Working Draft [6] for confirmation. Further, the use of file-scoped elements in C++, those declared global and static, is now discouraged by the conventional wisdom in favor of unnamed namespaces, which perform the same function more explicitly. In other words, each translation unit gets its own object for scoping, and the fact that the scope has no name means it can’t be used outside that translation unit. Again, objects are being used to scope and wrap elements that have previously been considered as freely roaming.
Listing 2.1. A simple method call as pseudocode.
1 类A {
B b;
3 f() {
bg();
5 };
};
7
类B {
9 g() {};
h() {};
11 };
13 main() {
A a;
15 af();
};
1 class A {
B b;
3 f() {
b.g();
5 };
};
7
class B {
9 g() {};
h() {};
11 };
13 main() {
A a;
15 a.f();
};
回到代码示例，a.f()依次调用b.g()，并且这两个方法之间存在关系。a和之间的关系f（封闭或作用域之一）是上下文相关的。 它有助于指定我们正在讨论的方法。b和之间存在类似的关系。 但是，从到 的g调用不仅仅是上下文相关的，它是这两个方法之间的主要关系。 换句话说，这是我们感兴趣的关系，但我们必须使用作用域来实现它。a.f()b.g()
Getting back to the code example, a.f() in turn calls b.g(), and a relationship exits between those two methods. The relationship between a and f, one of enclosure or scoping, is contextual. It helps specify which method we are talking about. A similar relationship exists between b and g. The call from a.f() to b.g(), however, is not just contextual, it is the primary relationship between those two methods. In other words, this is the relationship we’re interested in, but we have to use scoping to get there.
范围关系有助于我们完善对特定设计元素的视图。我们只需要两个这样的设计元素来形成 EDP 中描述的单一关系。可能有许多范围关系可以建立我们感兴趣的最终单一关系，但我们并不立即关注它们。它们是编程元素描述的一部分，这些元素构成了我们希望处理的关系的端点。回到我们之前关于房屋风格和设计的比喻，范围元素有点像说一块木头是一块 2×4 或半英寸厚的斜纹胶合板。这些信息并没有告诉我们如何将一个元素定位到另一个元素，也没有告诉我们在哪里钉钉子。范围提供了一个上下文来帮助定义元素是什么，但它并没有太多地解释该元素与系统中其他元素的关系。
Scoping relationships help us refine our view to a particular design element. We need exactly two such design elements to form the single relationship described in an EDP. There may be a number of scoping relationships that set up the final single relationship we are interested in, but we are not immediately concerned with them. They are part of the description of the elements of programming that comprise the endpoints of the relationship we wish to work with. Returning to our earlier metaphor of housing styles and design, the scoping elements are a bit like stating that a piece of wood is a two-by-four or a half-inch-thick sheet of bias-grain plywood. That information doesn’t tell us how to orient one to the other or where to drive the nails. Scoping provides a context to help define what an element is, but it doesn’t do much to explain how that element relates to others in the system.
现在，我们可以调整我们的问题，即我们是否已经达到了分解目标，将其改为，这是否体现了多个利益关系？这让我们离我们的目标更近了。现在我们只需要确定感兴趣的关系是什么。到目前为止，我们已经描述了范围关系，其中包括方法和字段的类所有权，以及命名空间、包和编程中可用的所有其他分组技术，并且我们表示我们目前对它们不感兴趣。让我们考虑一下我们可以在哪些剩余部分之间建立关系。
We can now tweak our question of whether we’ve reached our decomposition goal to, Does this embody more than one relationship of interest?, which gets us closer to our goal. Now we just have to determine what a relationship of interest is. So far, we’ve described scoping relationships, which include class ownership of methods and fields, as well as namespaces, packages, and all the other grouping techniques available in programming, and we stated that we’re not interested in them at the moment. Let’s consider instead what remains that we could form relationships between.
剩下的就是类、它们的字段和方法，仅此而已。列表中似乎缺少一项：对象。毕竟，这是面向对象编程！为了完整起见，我们将把对象添加到列表中，稍后我们将展示它们对这种方法的核心作用。（有关正式解释，请参阅附录。）目前，我们的列表中有这四种编程实体：对象、方法、字段和类或类型。这似乎是一件奇怪的事情，但实际上却很有道理。
What remains are classes, their fields and methods, and little else. One item seems to be missing from the list: objects. It is object-oriented programming, after all! We’ll add objects to the list for completeness, and later we’ll show how they’re central to this approach. (For a formal explanation, see the appendix.) For now, we have these four kinds of programmatic entities on our list: objects, methods, fields, and classes, or types. This may seem like an odd thing to do, but it actually makes solid sense.
我们不称它们为类，因为并非每种面向对象语言都有类，但每种面向对象语言都有类型。请记住，我们希望以一种能够跨越语言界限的方式来描述设计模式，因此我们将寻找每种面向对象语言所共有和要求的功能。那么，我们如何从大多数人熟悉的类过渡到更经典的“纯”面向对象方法呢？其实这并不复杂：只需将类拆分为其组成部分即可。
We don’t call them classes, because not every object-oriented language has classes, but every object-oriented language has types. Remember that we’re looking to describe design patterns in a way that can cross language borders, so we’re going to be looking for features that are common to, and a requirement for, each and every object-oriented language. So how do we make the transition from classes, which most people are familiar with, to a more classically “pure” object-oriented approach? It’s not that complex, really: it simply involves splitting a class into its constituent parts.
类是一种有趣的东西，它在大多数面向对象语言中非常常见，以至于许多学生和开发人员认为它是主要和必要的元素。但从严格的面向对象的角度来看，事实并非如此。有些语言，如 Self 和 Lua，甚至没有类的概念。相反，它们依靠原型设计、克隆和对对象的其他操作来执行相同的功能。在大多数语言使用类的地方，不仅仅是深奥的语言展示了对象的这种用法，因为在 JavaScript 中可以看到这种用法的各种形式。即使是 Smalltalk（可以说是大多数面向对象语言的祖先），尽管使用了相同的术语，但对类的实现和理解也大不相同。
A class is an interesting beast in that it is so common in most object-oriented languages that many students and developers consider it a primary and necessary element. The reality, from a strictly object-oriented viewpoint, is that it is not. Some languages, such as Self and Lua, don’t even have the concept of a class. Instead, they rely on prototyping, cloning, and other actions on objects to perform the same functions. It’s not just esoteric languages that exhibit this use of objects in places where most languages use classes, as a variety of this usage can be seen in JavaScript. Even Smalltalk, arguably the progenitor of most object-oriented languages, has a quite different implementation and understanding of a class even though the same term is used.
一般来说，当前常见的面向对象语言（如 C++ 和 Java）中的类执行两项任务。首先，它描述该类实例化对象中的元素、成员方法和字段。我们大多数人都希望以这种方式使用类。其次，它描述该类所有实例化对象中共有的元素、类方法和字段。在 C++ 或 Java 中，当我们将方法或字段声明为时，就会发生这种情况static。第一个用例在形式上直接对应于类型，第二个用例可以可以使用特殊用途的对象重新创建。考虑 C++ 类中声明为静态的字段及其访问方式，如清单 2.2所示。
In general, a class in current common object-oriented languages such as C++ and Java performs two duties. First, it describes the elements that will be in instantiated objects of that class, the member methods and fields. Most of us expect a class to be used in this way. Second, it describes elements that are common across all instantiated objects of that class, the class methods and fields. This is what happens when, in C++ or Java, we declare a method or field to be static. The first use case corresponds directly to a type in the formal sense, and the second use case can be re-created using a special-purpose object. Consider a field declared static in a class in C++ and how it is accessed, as in Listing 2.2.
列表 2.2. 类、实例和命名空间内的字段，如 C++ 中定义和使用的。
Listing 2.2. Fields within classes, instances, and namespaces, as defined and used in C++.
命名空间MyNamespace {
2 int aField;
};
4
class MyClass {
6 public:
static int sharedField;
8 int instanceField;
};
10
main() {
12 MyClass mc;
mc.instanceField = 0;
14 MyClass::sharedField = 1;
MyNamespace::aField = 2;
16 };
namespace MyNamespace {
2 int aField;
};
4
class MyClass {
6 public:
static int sharedField;
8 int instanceField;
};
10
main() {
12 MyClass mc;
mc.instanceField = 0;
14 MyClass::sharedField = 1;
MyNamespace::aField = 2;
16 };
请注意，在main使用类字段时sharedField，类名 是放在前面的。这与在已定义 中MyClass访问时使用的符号完全相同。我们可以将保存类的类共享元素的实体视为“活动”对象，也可以以同样的方式考虑命名空间。我们提供对象的名称以及我们想要从中选择的部分，就像以正常方式从类中实例化的任何对象一样。当语言提供命名空间、包或类时，它就提供了一种自动实例化这些对象的方法。aFieldmyNamespace
Notice that in main the use of the class field sharedField is prefaced by the class name MyClass. This is exactly the same notation used to access aField in the defined myNamespace. We can think of the entity that holds the class-shared elements of a class as a “live” object, and we can think of a namespace in the same way. We provide the name of the object and the piece we want selected out of it, just like any object that was instantiated from a class in the normal way. When the language provides a namespace, a package, or a class, it is providing a way of instantiating these objects automatically.
还不相信吗？在 Smalltalk 中，类由对象表示。从字面上讲，它被称为类对象。要实例化该类的对象，我们向类对象发送一条消息（相当于 Smalltalk 中的“调用对象的方法”）以返回该类的实例。类方法和字段存在于该类对象中。这几乎就是我们在此处描述的场景。
Still not convinced? In Smalltalk, the class is represented by an object. Literally, it is named the class object. To instantiate an object of that class, we send the class object a message—the Smalltalk equivalent for “call a method of the object”—to return an instance of that class. Class methods and fields live in that class object. It is almost exactly the scenario we are describing here.
或者，看看 Java。在 Java 中，类对象是一种特殊对象，可通过反射进行检查，并与开发人员定义的类相关联。与 Smalltalk 不同，开发人员在“正常”使用中不会直接访问它，但通常保留用于更复杂的反射操作。与 Smalltalk 中相同的基本原则适用，只是从new MyClassC++ 借用的结构将过程隐藏在了一些语法糖之下。
Or, look at Java. In Java the class object is a special object that is available for inspection through reflection and is associated with a class that a developer defines. Unlike in Smalltalk, it is not directly accessed by the developer in “normal” use but is usually reserved for more complex reflection actions. The same basic principle applies as in Smalltalk, except that the new MyClass construct borrowed from C++ hides the process under a bit of syntactic sugar.
清单 2.3 的上半部分显示了 Java 代码片段，下半部分显示了该类的一种可能的分解。如果您将通常的 视为new MyClass的同义词（MyClass_Object.new()在本例中），然后通过 访问静态成员MyClass_Object，如下所示MyClass_Object.sharedData或者MyClass_Object.sharedMethod()，那么它与您可能已经习惯的 Java 没什么不同。C++ 也有类似的机制，但没有反射选项。
Listing 2.3 shows a Java snippet in the top half and one possible decomposition of the class in the bottom half. If you think of the usual new MyClass as being a synonym for MyClass_Object.new() in this case, and then access to the static members being done through the MyClass_Object instead, as in MyClass_Object.sharedData or MyClass_Object.sharedMethod(), then it’s not that different than what you may be used to in Java already. Similar mechanisms are available for C++, where no reflection option exists.
列表 2.3。一个 Java 类以及一个可能的等效对象和类型。
Listing 2.3. A Java class, and one possible equivalent object and type.
类 MyClass {
2 公共：
静态 int sharedData;
4 int instanceData;
静态 void sharedMethod() { ... };
6 void instanceMethod() { ... };
};
8
10
MyClass：
12 int instanceData;
void instanceMethod();
14
MyClass_Object：
16 int sharedData;
void sharedMethod();
18 MyClass new ();
class MyClass {
2 public:
static int sharedData;
4 int instanceData;
static void sharedMethod() { ... };
6 void instanceMethod() { ... };
};
8
10
MyClass:
12 int instanceData;
void instanceMethod();
14
MyClass_Object:
16 int sharedData;
void sharedMethod();
18 MyClass new();
尽管处理细节因语言而异，但在每种情况下，都可以使用类型和对象来模拟类。虽然乍一听可能有点陌生，但在实践中很快就会习以为常。只需记住，没有静态成员的类（使用 C++ 和 Java 术语）只是一种类型。任何定义的类成员都将移至相应的类对象中，如在 Smalltalk 中，或移至反射访问的 Java 类对象中。
Although the details of how it is handled will differ from language to language, in every case a class can be emulated using a type and an object. While this may sound a bit foreign to you at first, it quickly becomes second nature in practice. Just remember that a class with no static members (using the C++ and Java terminology) is just a type. Any defined class members would be moved into a corresponding class object, as in Smalltalk, or into the reflection-accessed Java class object.
我们需要类型来满足类型化语言的基本属性，而对象是面向对象语言的核心，无论其类特征如何。通过结合使用这两者，我们可以模拟基于类的面向对象语言中的类。这使我们能够将定义设计模式所需的项目集简化为前面提到的四个项目：对象、方法、字段和类型。
We needed types in order to satisfy basic properties of typed languages, and objects are the core of object-oriented languages regardless of their class features. By using these two in conjunction, we can emulate classes from class-based object-oriented languages. This allows us to simplify our set of items needed to define design patterns to just our four previously mentioned items: objects, methods, fields, and types.
这个简短列表中的项目可以以哪些方式交互？表 2.2是左列中的实体与顶行中的实体交互的可能方式的完整列表。对象和类型可以包含并因此定义这四个元素中的任何一个，但我们将从作用域机制的考虑中消除这些“包含”关系。在某些语言中，方法可以类似地定义或作用域内部方法或类型，当然，我们可以定义局部变量或字段。方法还可以调用其他方法，使用非本地字段，并具有返回类型。字段更简单，因为它们可以被分配方法或另一个字段的返回值。当然，它们有一个类型。最后，类型可以通过范围定义四种实体类型中的任何一种，并且类型可以通过子类型依赖于其他类型。
In what ways can the items on this short list interact? Table 2.2 is a complete list of possible ways in which the entities in the left column can interact with the entities on the top row. Objects and types can contain, and therefore define, any of the four elements, but we’re going to eliminate those “contains” relationships from consideration under the scoping mechanism. Methods can, in some languages, similarly define, or scope, inner methods or types, and of course we can define local variables or fields. Methods can also call other methods, use nonlocal fields, and have a return type. Fields are simpler in that they can be assigned the return value of a method or of another field. And, of course, they have a type. Finally, types can define through scope any of the four entity kinds, and types can rely on other types through subtyping.
Table 2.2. All interactions between entities of object-oriented programming
现在可能实现的交互并不多，这意味着我们可以开始将它们枚举为一个有限的、更好的是相当小的可能性集。表 2.3显示了不包括定义另一个实体的交互。
Not that many interactions are possible now, which means that we can start enumerating them into a finite and, better yet, quite small set of possibilities. Table 2.3 shows the interactions that don’t include defining another entity.
Table 2.3. Nonscoping interactions between entities of object-oriented programming
字段具有其所依赖的类型；这只是字段的类型。对象也是如此。同样，方法也有返回类型。它们用于定义一段数据是什么，无论它是字段还是另一个对象（例如参数或方法的返回值）。与范围非常相似，定义数据的感觉是描述对象或字段本身，而不是提供两个实体之间的关系。
A field has a type on which it relies; this is simply the type of the field. The same holds for objects. Similarly, methods have return types. They are used to define what a piece of data is, whether it is a field or another object such as a parameter or a return value from a method. Much like scoping, defining data has the feel of describing the object or field itself, not providing a relationship between two entities.
例如，查看清单 2.4中的代码片段。数据成员pos被定义为类型的实例Position。这告诉我们pos 是什么，并且其范围Glyph告诉我们如何解释它，但它并没有告诉我们它如何pos与系统中的其他元素交互。同样，方法参数的类型scaleCopy只告诉我们这些参数是什么，而不是它们与其他任何东西的关系。
For instance, look at the code snippet in Listing 2.4. The data member pos is defined as being an instance of the type Position. This tells us what pos is, and its scoping within Glyph tells us some of how to interpret it, but it doesn’t tell us how pos interacts with other elements in the system. Similarly, the types of the parameters to the scaleCopy method only tell us what those parameters are, not how they relate to anything else.
另一方面，其他关系（例如，一种类型依赖于另一种类型）提供的信息比仅仅描述一个元素要多。类型依赖其他类型，最常被视为继承，有相当多的记录。我们甚至可以说它是根深蒂固和易于理解的。
On the other hand, other relationships, such as a type relying on another type, provide more information than just a description of one element. Type reliance on other types, which is most often seen as inheritance, is rather well documented. We might even say it is ingrained and understood.
Listing 2.4. Typing as context.
类Glyph {
2 位置 pos ;
字形 scaleCopy( float x, float y ) ;
4 }
class Glyph {
2 Position pos;
Glyph scaleCopy( float x, float y );
4 }
为了定义 EDP，表 2.3中间只剩下四种基本关系，其中方法或字段依赖另一个方法或字段来获取其值、完成其工作等等。这些关系是方法调用（方法依赖于方法）、方法使用的字段（方法依赖于字段）、方法设置的字段（字段依赖于方法进行状态更改）以及字段依赖于另一个字段，例如a = b + 1。如前面的表所示，我们可以将它们称为方法调用、字段使用、状态更改和内聚性。信不信由你，几乎整本书都集中在第一个方面，即方法调用依赖。是的，如果从正确的角度来看，简单的方法调用依赖有很多话要说。
For the purpose of defining the EDPs, there are only the four basic relationships left in the center of Table 2.3, where either a method or a field relies on another method or field for its value, to get its job done, and so on. Those relationships are a method call (method relies on method), a field use by a method (method relies on a field), a field being set by a method (field relies on method for a state change), and a field relying on another field, such as a = b + 1. As shown in the preceding tables, we can call these a method call, a field use, a state change, and cohesion. Believe it or not, nearly this entire book concentrates on just the first one, a method call reliance. Yes, there is a lot to be said about a simple method call reliance when looked at in the right light.
您可能已经想到，方法调用看起来非常像结构实体，而不是概念实体。这是事实，这就是我之前开始偷偷使用依赖这个词的原因。一个元素对另一个元素的依赖不需要直接联系。考虑一个在其主体中f调用方法的方法。我们说依赖于。现在假设反过来调用另一个方法，如清单 2.5中所示。我们说依赖于。这种关系是传递的应该是不言而喻的 - 如果依赖于，并且依赖于，那么也依赖于。这不是直接依赖，但这并不重要：仍然依赖于完成其工作，以便可以完成其工作。gfgghghfgghfhfhf
It has probably occurred to you that a method call looks an awful lot like a structural entity, not a conceptual one. This is true, and that’s why I started sneaking in the word reliance earlier. A reliance by one element on another doesn’t require a direct connection. Consider a method f that calls method g in its body. We say that f relies on g. Now assume that g in turn calls another method h, as in Listing 2.5. We say that g relies on h. It should be self-evident that this relationship is transitive—if f relies on g, and g relies on h, then f also relies on h. It’s not a direct reliance, but that doesn’t really matter: f still relies on h to complete its work so that f can do its work.
让我们稍微扭转一下。如果我们想要f依赖h，那么它是直接方法调用还是方法g位于中间重要吗？这并不重要。只要通过某种路径f依赖于h，我们的要求就成立。我们刚刚从结构连接（方法调用）跳跃到了概念连接（方法调用依赖）。这使我们不必以结构化的方式讨论设计的低级编程概念，而使我们能够以概念性的方式描述它们，如前所述，这对于使用 SPQR 和查找设计模式实例是必要的。我们将在第 4 章第4.1.1 节中更详细地讨论这一点，并展示它如何成为使用 EDP 的关键部分。
Let’s turn this around a bit. If we want f to rely on h, does it matter if it is a direct method call or if method g is in the middle? It does not. As long as f relies on h by some path, our requirement holds true. We just made the leap from a structural connection, that of a method call, to a conceptual connection, that of a method call reliance. This frees us from having to discuss low-level programming concepts of design in a structural manner and enables us to describe them in the conceptual way outlined earlier as necessary for working with and finding design pattern instances with SPQR. We discuss this in more detail and show how it forms a critical portion of working with EDPs in Chapter 4, Section 4.1.1.
在教授 SPQR 编程概念的过程中，我们摆脱了僵硬的编程结构，转而获得了灵活且宽容的方式来描述构成软件设计的关系。通过我们自己使用这些相同的技术，我们为开发人员和设计人员提供了同样灵活的思维和能力，以有条不紊但易于理解的方式将设计从实现中抽象出来。我们用同样的方法实现了双重任务。
In the process of teaching SPQR about programming concepts, we made the transition away from rigid structures of programming and instead gained a flexible and forgiving way to describe the relationships that form software design. By using these same techniques ourselves, we provide developers and designers with the same flexible thinking and ability to abstract design from implementation in a methodical yet understandable manner. We get double duty out of the same approach.
Listing 2.5. A method call chain as pseudocode.
函数f() {
2 g();
};
4
函数g() {
6 h();
};
8
函数h() {};
10
函数main {
12 f();
};
function f() {
2 g();
};
4
function g() {
6 h();
};
8
function h() {};
10
function main {
12 f();
};
让我们回顾一下。我们将编程中有趣的元素列表缩减为四个：对象、方法、字段和类型（根据实现语言的不同，您可以将其视为类）。其他编程实体（例如命名空间、包等）可帮助我们描述正在查看的元素，但它们不会创建我们在检查非平凡设计时关注的那种关系。在这四个元素中，我们最关注的是四种关系或依赖关系：方法-方法、方法-字段、字段-方法和字段-字段。在本书中，我们几乎只关注方法-方法依赖关系。
So let’s recap. We have reduced our list of interesting elements of programming to just four: objects, methods, fields, and types (which, depending on your implementation language, you may think of as classes). Other programmatic entities, such as namespaces, packages, and so on, help us describe which element we’re looking at, but they don’t create the kind of relationships we’re focused on when inspecting nontrivial design. Of those four elements, there are really four relationships, or reliances, that we’re most concerned with: method–method, method– field, field–method, and field–field. We concentrate almost exclusively on the method–method reliance in this book.
这种仅基于方法调用的依赖真的能帮助我们描述更大的设计模式文献吗？只要有正确的背景，一切皆有可能。
Can this one reliance, based solely on method calls, really do anything useful to help us describe the larger design pattern literature? Given the right context, anything is possible.
2.2.3. 背景
2.2.3. Context
上一节主要讲的是为我们的讨论找到正确的关系。我们重点关注了主要依赖关系，例如方法和字段之间的依赖关系，而忽略了其他类型的关系，例如类型依赖或继承。现在是时候将这些上下文关系重新纳入讨论中了。
The previous section was all about finding the right relationships for our discussion. We focused on primary reliances such as those between methods and fields, bypassing other kinds of relationships, such as type–type reliances or inheritance. Now it’s time to bring these contextual relationships back into the discussion.
对于给定的方法调用依赖，我们可以使用另外三条信息来帮助我们弄清楚该依赖在特定设计中的目的是什么。如果您查看方法调用，这些信息就很明显，如下所示图 2.5 . 在面向对象理论中，实际上没有所谓的普通方法或函数。这并不罕见。在混合面向对象编程语言（如 C++）中，可以将函数放在全局空间中。在更纯粹的语言（如 Java）中，则不能这样做；每个方法都必须位于一个对象中，要么作为类的实例，要么作为静态类级方法，这相当于将方法与该类的类对象放在一起。让我们假设每个方法都必须嵌入在一个对象中。对于任何给定的方法调用依赖，都有四个部分：调用方法、被调用方法和每个方法所在的对象（类对象或实例）。清单 2.6在 C++ 中设置了此类方法调用的类和对象。图 2.5说明了方法of()调用方法pg()的四个部分。这些部分存在于每个方法调用中。
For a given method call reliance, there are three other pieces of information we can work with to help us figure out what the purpose of that reliance is in a particular design. These are obvious if you look at a method call, as shown in Figure 2.5. In object-oriented theory, there’s really no such thing as a plain method or function. This isn’t that unusual. In hybrid object-oriented programming languages such as C++, you can put functions in the global space. In more pure languages such as Java, you cannot; each method must be in an object, either as an instantiation of a class or as a static class-level method, which is equivalent to placing the method with the class-object for that class. Let’s assume that every method must be embedded within an object. For any given method call reliance, then, there are four pieces: the calling method, the called method, and the object (class-object or instance) each resides in. Listing 2.6 sets up the classes and objects for such a method call in C++. Figure 2.5 illustrates the four pieces of the method o.f() calling the method p.g(). These pieces are present in every method call.
图 2.5。方法调用的各个部分。
Figure 2.5. The parts of a method call.
清单 2.6.图 2.5的简单方法调用。
Listing 2.6. Simple method call for Figure 2.5.
1 类Class2 {
void g() {};
3 };
5 类Class1 {
void f() {
7 Class2 p;
pg();
9 }
};
11
void main() {
13 Class1 o;
of();
15 };
1 class Class2 {
void g() {};
3 };
5 class Class1 {
void f() {
7 Class2 p;
p.g();
9 }
};
11
void main() {
13 Class1 o;
o.f();
15 };
显而易见的三条信息是：
The three pieces of information hiding in plain sight are:
1.包围对象之间的相似性
1. The similarity between the enclosing objects
2.包围对象类型之间的相似性
2. The similarity between the types of the enclosing objects The
3.调用方法与被调用方法的相似性
3. similarity between the calling method and the called method
我在这里引入了一个新词：相似性。通俗地说，它的意思就是你认为的两件事之间的相似性。但在方法调用依赖的上下文中，它是什么意思呢？
I’ve introduced a new word here: similarity. Colloquially, it means what you think—a resemblance between two things. But what does it mean in the context of method call reliance?
对象相似度是指一个对象与另一个对象的相似程度。它们是同一个对象吗？它是通过指针指向第二个对象的别名吗？它们是完全不相关的吗？
Object similarity is the extent to which one object is like another. Is it the same object? Is it an alias to the second object through a pointer? Is it completely unrelated?
我们还可以讨论对象类型之间关系的相似性。它们是同一类型吗？一个是另一个的子类型吗？或者一个是另一个的兄弟类型，具有共同的超类型祖先？
We can also discuss the similarity of relationship between the types of the objects. Are they the same type? Is one a subtype of the other? Or maybe one is a sibling type of the other, with a common supertype ancestor?
方法相似性有点棘手。我们需要确定两种方法是否试图执行类似的任务。我们可以查看许多方面：我们可以映射注释中的关键字，或者对方法主体进行全面分析以确定它们在计算上完成了什么。但是，如果我们借鉴社会工程学，查看方法名称，并在较小程度上查看方法签名，那么还有一种更简单的方法。
Method similarity is a bit trickier. We need to determine whether two methods are trying to do a similar task. We could look at many aspects: we could map keywords in comments, perhaps, or do a full analysis of the method bodies to determine what they computationally accomplish. There is a much easier way, however, if we take a page from social engineering and look at the method names and, to a lesser extent, the method signatures.
不，真的。想一想。开发人员传达方法意图的主要方式是什么？名称。这甚至隐藏在 Kent Beck 的 [ 5 ]意图揭示选择器模式中（重点是我加的）：“在命名方法时，你有两种选择。第一种是根据方法完成任务的方式命名方法。……反对这种命名方式的最重要论点是它不能很好地传达信息。……第二种选择是根据方法应该完成的任务命名，把“如何”留给各种方法体。这是艰苦的工作，特别是当你只有一个实现时。你的脑海里充满了你将如何完成任务，所以名字自然会跟在“如何”后面。将方法的名称从“如何”改为“什么”的努力是值得的，无论从长期还是短期来看。最终的代码将更易于阅读，也更灵活。”Beck 总结道：“根据方法完成的任务来命名方法。”
No, really. Think about it for a moment. What’s the primary means by which a developer conveys the intent of a method? The name. This is even ensconced in Kent Beck’s [5] Intention Revealing Selector pattern (emphasis mine): “You have two options in naming methods. The first is to name the method after how it accomplishes its task. . . . The most important argument against this style of naming is that it doesn’t communicate well. . . . The second option is to name a method after what it is supposed to accomplish and leave “how” to the various method bodies. This is hard work, especially when you only have a single implementation. Your mind is filled with how you are about to accomplish the task, so it’s natural that the name follow how. The effort of moving the names of method from “how” to “what” is worth it, both long term and short term. The resulting code will be easier to read and more flexible.” Beck sums this up as, “Name methods after what they accomplish.”
罗伯特·马丁 (Robert Martin) 的《代码整洁之道》 [ 27 ] 用整整一章来讲述命名，因为它太重要了。也许可以用书中引用的一句老话来最好地概括这一点：“说出你的意思。说到做到。”这只是一种很好的做法，虽然不是每个开发人员团队都会虔诚地遵循它，但它足够普遍，可以利用它来为我们省去大量的辛苦工作。在最坏的情况下，我们可以依靠同义词查找和单词重新排序来实现这一点，makeAString并且stringCreator意图接近，但令人惊讶的是，在大多数情况下，一个简单的词典顺序比较就能达到目的。
Robert Martin’s Clean Code [27] spends an entire chapter on naming alone, because it’s so important. Perhaps this can be best encapsulated using that text’s invocation of the old adage “Say what you mean. Mean what you say.” This is just good practice, and although not every developer group will follow it religiously, it is common enough that it can be leveraged to save us a tremendous amount of hard work. At worst, we can fall back on synonym lookups and word reordering for realizing that makeAString and stringCreator are close in intent, but it’s rather shocking how often a simple lexicographic comparison will do in most cases.
重载方法（即具有相同名称但不同参数列表的方法）会带来一些小问题，但同样，我们可以看看人类开发人员如何推断相似性以寻找线索。与名称完全不同的方法相比，具有相同名称但不同参数列表的方法更相似。完美匹配是具有相同名称和相同参数列表的方法。不过，出于本文的目的，我们可以先不讨论这个细节。
Overloaded methods, those with the same name but different argument lists, offer a small hiccup, but again, we can look to how human developers reason about similarity for clues. A method is more similar to one with the same name but a different argument list than it is to a method with a completely different name. A perfect match is one with the same name and the same argument list. For the purposes of this text, however, we can set aside this detail.
SPQR 使用的就是这种算法，其结果令人惊叹。我最初选择该算法仅仅是因为它最容易实现，我从未指望它真的能起作用。它只是一个占位符，直到经过初步测试后可以选出合适的方法。一旦 SPQR 开始生成数据，并且很明显这种方法在大多数情况下都足够了，就必须找出原因。回想起来很明显，但事后看来，意外的发现通常更清楚。
This algorithm is what SPQR uses with stunningly good results. I originally selected the algorithm solely because it was the simplest to implement, and I never expected it to actually work. It was to be only a placeholder until an appropriate approach could be selected after initial testing. Once SPQR started producing data and it became obvious that this method was sufficient for most cases, the reason why had to be worked out. It’s obvious in retrospect, but then, unexpected discoveries usually are clearer in hindsight.
我们不应该对这种方法的有效性感到惊讶。当开发人员阅读代码以确定其用途时，他或她做的第一件事就是查看事物的名称，将其作为其用途的线索。这就是 Beck 命名原则的作用——我们希望事物的命名恰当，更重要的是，命名一致。我们希望执行类似功能的事物具有类似的名称。我们在这里所做的就是利用这种期望。SPQR 只是被教导做我们在阅读代码时所做的同样的事情：尝试确定事物之间的关系。反过来，这种经验指导了 EDP 目录的形成，它提供了对人类如何执行该任务的更好理解，并提供了一种简单而自然的方式来描述我们的第三个相似性。
We shouldn’t have been surprised that this approach worked. When a developer reads code to try to ascertain its purpose, the first thing he or she does is look at the names of things as clues to what they do. This is Beck’s naming principle at work– we expect things to be named appropriately, and more to the point, consistently. We expect things that perform similar functions to have similar names. All we’re doing here is leveraging that expectation. SPQR was simply taught to do the same thing we all do when reading code: to try to ascertain how things relate to each other. In turn, that experience guided the formation of the EDP catalog by providing a better understanding of how humans perform that task and offering a simple and natural way to describe our third similarity.
因此，对于任何方法调用，我们都可以通过一些相当简单的分析（无论是自动分析还是手动分析）来查看所涉及的对象、这些对象的类型以及方法的相似性。这给我们带来了什么？
For any method call, we can therefore look at the objects involved, the types of those objects, and the similarity of the methods through some fairly simple analysis, whether automated or manual. What does this give us?
它为我们提供了三个独立的上下文轴，可以在其中放置和描述任何方法调用。我们可以将这三个轴视为定义一个我们可以开始探索的三维空间。根据我们三个上下文的相似性评估，任何方法调用都将位于这个空间中的某个位置。换句话说，共享同一空间的所有方法调用依赖关系都具有相似的形式，我们可以独立于编写它们的人、编写方式、它们出现的软件系统甚至用于实现它们的语言来描述这些形式。
It gives us three independent axes of context in which to place and describe any method call. We can think of these three axes as defining a three-dimensional space that we can start to explore. Any method call will, depending on the similarity assessments for our three contexts, sit in exactly one spot within this space. In other words, all method call reliances that share the same space have a similar form, and we can describe those forms independently of who wrote them, how they were written, the software system they appear in, or even the language used to implement them.
我们可以根据依赖项的作用以及使用它们的目的来描述它们。我们可以讨论何时使用它们效果最好、它们与三维空间中周围的其他依赖项有何关系以及如何从一种方法调用依赖项转换为另一种方法调用依赖项。
We can describe the reliances on the basis of what they do for us and what we intend when we use them. We can discuss when they’re best used, how they relate to other reliances around them in that three-dimensional space, and how to transform from one method-call reliance to another.
这给了我们第一组 EDP。
This gives us the first group of the EDPs.
2.2.4. 设计空间
2.2.4. The Design Space
继续依赖方法调用，让我们探索一下这个三轴定义的空间可能是什么样子。我们有三种正交且独立的方式根据三种关系描述单个方法调用：包含方法的对象、这些对象的类型以及方法本身。
Sticking with our method call reliance, let’s explore what this three-axis-defined space might look like. We have three orthogonal and independent ways of describing a single method call based on three relationships: the objects that enclose the methods, the types of those objects, and the methods themselves.
这看起来似乎有很多东西需要记住，但很容易形象化。为了简单起见，我们可以从方法调用开始，忽略对象类型关系，而是专注于对象和方法的相似性，如图2.6所示。
This may seem like a lot to remember, but it’s easy enough to visualize. For simplicity, we can start with method calls in which we ignore the object type relationship, instead concentrating on the object and method similarities, as in Figure 2.6.
图 2.6.一个简单的设计空间。
Figure 2.6. A simple design space.
我们有一个三乘三的网格，沿着每个轴，我们分别定位相似、未知和不相似。包含未知的原因将在稍后解释。很明显，中心方块只是一个普通的方法调用。我们对所涉及的对象或方法一无所知。令人惊讶的是，这是大多数软件分析和研究的重点。当有人谈论方法调用而不指出方法、方法所包含的对象或这些对象的类型之间的关系时，那么所有这些方法调用都存在于那个方块中。所有方法调用。
We have a three-by-three grid and, along each axis, positions for similar, unknown, and dissimilar. Unknown is included for reasons that will be explained shortly. It should be pretty obvious that the center square is just a plain, vanilla method call. We don’t know anything about the objects or the methods involved. Surprisingly, this is where most software analysis and research lives. When someone talks about method calls without indicating a relationship between the methods, their enclosing objects, or the types of those objects, then all of those method calls live in that one square. All of them.
在顶部中心方块中，我们有具有对象相似性的方法调用。为了便于讨论，我们暂且称这种相似性为对象等价性。这些是单个对象内的方法调用次数，称为对象的内聚力，或者对象与自身方法的绑定程度[ 42 ]。
In the top center square, we have method calls with object similarity. For the purposes of this discussion, let’s just call that similarity object equivalence. These are method calls within a single object, known as an object’s cohesion, or how tightly it is bound with its own methods [42].
在底部中心的方块中，我们有不同的对象，或者就本讨论而言，不同的对象实例，方法调用构成了耦合的基础，耦合是对象之间绑定的度量[ 15 ]。
In the bottom center square, where we have dissimilar objects, or for this discussion, different object instances, method calls form the basis for coupling, a measure of the binding between objects [15].
顶部和底部中心方块共同代表了大量研究的核心 [ 7 – 9 , 13 , 16 , 22 , 23 , 31 , 34 ]，它们与原始中心方块一起构成了我们定义的空间的三分之一。另外三分之二是什么？最有趣的是四个角落中的那些位置，我们在那里了解对象和方法。
The top and bottom center squares together represent the core of a wide body of research [7–9,13,16,22,23,31,34], which, along with the original center square, form just one-third of the space we’ve defined. What is in the other two-thirds? The most interesting slots are those in the four corners, where we have knowledge of both the objects and the methods.
图 2.7用与之相关的编程概念的名称填充了这四个角。例如，一个对象对自身的方法调用，也就是我们所知的“递归” 。我们可以像清单 2.7中那样编写它，其中递归发生在内部countDown()。这一点很简单也很明显，但其他三个呢？
Figure 2.7 fills out these four corners with the names of the programming concepts associated with them. For instance, a method call from an object to itself, to the same method, is what we all know as Recursion. We can write this as in Listing 2.7, with recursion occurring within countDown(). That one’s easy and obvious, but what about the other three?
图 2.7.带有 EDP 的简单设计空间。
Figure 2.7. A simple design space with EDPs.
Listing 2.7. Example of a Recursion method call in Java.
1 class Timer {
public void countDown( int counter) {
3 if (counter > 0) {
this .countDown(counter--);
5 } else {
// 暂时忽略此分支
7 }
};
9 };
1 class Timer {
public void countDown(int counter) {
3 if (counter > 0) {
this.countDown(counter--);
5 } else {
// Ignore this branch for now
7 }
};
9 };
Listing 2.8. Example of a Delegation method call in C++.
1 class VicePresidentOfSales {
public:
3 void increaseQuarterlySales();
};
5
class CEO {
7 VicePresidentOfSales vpOfSales;
void increaseProfits() {
9 vpOfSales.increaseQuarterlySales();
};
11 };
1 class VicePresidentOfSales {
public:
3 void increaseQuarterlySales();
};
5
class CEO {
7 VicePresidentOfSales vpOfSales;
void increaseProfits() {
9 vpOfSales.increaseQuarterlySales();
};
11 };
在对角，我们看到了委托，这是一个包罗万象的词，在软件设计中以多种方式使用。在这里，我们给它一个精确的定义，表示两个不同对象之间以及不同方法之间的方法调用。清单 2.8显示了一个例子。为什么要使用委托这个词？因为调用方法将其部分工作委托给另一个对象中的另一个方法。它的意思是，“去做这个，这样我就能完成我的工作了。”它分配了一部分工作，这些工作是其自身预期任务的一部分，但与自身预期任务不同。想象一下一家公司的首席执行官将工作委托给负责各个部门的副总裁。首席执行官的职责是“经营公司”，但每个副总裁的职责描述不同：“管理人力资源”、“确保财务忠诚度”、“调查和部署 IT 技术”。首席执行官要求每个副总裁做一项看起来与他自己的任务完全不同的任务，但每一项任务对于首席执行官成功完成工作都至关重要。
In the opposite corner, we see Delegation, which is a catch-all word used in a number of ways in software design. Here we give it a precise definition to mean a method call between two distinct objects and between dissimilar methods. An example is shown in Listing 2.8. Why use the word delegation? Because the calling method is delegating part of its work to another method in another object. It’s saying, “Go do this so I can get my work done.” It’s doling out a portion of its work that is part of, but unlike, its own intended task. Think about the CEO of a company delegating to the vice presidents in charge of various departments. The CEO has a job described as “run the company,” but each VP has a job described in different ways: “manage HR,” “ensure financial fidelity,” “investigate and deploy technologies in IT.” The CEO is asking each VP to do a task that looks nothing like his own, but each is critical to the CEO successfully completing his job.
将其与右下角标记为“重定向”的图进行对比。调用方法现在要求另一个对象执行一些工作，但它要求完成的方法本身应该完成的相同工作。它将部分工作量重定向到另一个对象，但那部分看起来很像它自己的整体任务。清单 2.9中显示了一个示例。这有点像一条装配线在执行特定任务期间分支到多个单元并行工作，然后将结果重新集中到一条主线上继续到下一个站点。想象一下您的公司正在生产汽车，您负责喷漆。但是，由于干燥时间较长，如果您按顺序自己做的话，就会耽误其他所有人的时间。您不是一次只做一件事，而是雇用几个团队并为每个团队设置喷漆站。现在，您负责在他们完成前一辆汽车后将汽车分配给每一辆。您的工作是确保汽车喷漆完毕。他们的工作是确保汽车喷漆完毕。他们的工作是您工作的一个子集，看起来非常相似。您承担了分配责任，但这只是完成主要任务的一部分：确保汽车喷漆完毕。分配是一个实施细节，您与下属之间工作上的相似性使这种工作分配成为重定向而不是委派。重定向和委派共同构成了对象耦合的基础。
Contrast this with the bottom right corner, labeled Redirection. The calling method is now asking another object to do some work, but it is asking for the same work to be done that the method itself is supposed to do. It is redirecting part of its workload to another object, but that part looks a lot like its own overall task. An example is shown in Listing 2.9. This is a bit like an assembly line branching out to work on multiple units in parallel during a particular task, then bringing the results back together onto one main line to continue to the next station. Imagine that your company is building cars, and you’re in charge of painting them. The long drying time, however, means that you would be holding up everyone else if you did them yourself in sequence. Instead of doing them all yourself, one at a time, you hire several teams and set up paint stations for each of them. Now you’re in charge of routing the cars to each one as they complete the previous one. Your job is to ensure the cars are painted. Their jobs are to ensure the cars are painted. Their job is a subset of yours that looks awfully similar. You’ve taken on a routing responsibility, but that’s only as part of completing your primary task: ensure the cars are painted. The routing is an implementation detail, the similarity in work between you and your subordinates is what makes this type of parceling out of work redirecting as opposed to delegating. Together, Redirection and Delegation form the basics of how objects can be coupled.
清单 2.9。Objective -C 中的重定向方法调用示例。
Listing 2.9. Example of a Redirection method call in Objective-C.
1 @interface Painter {}
-( void ) paintCar: (Car) theCar;
3 @end
5 @interface PaintShopManager {
Painter subpainter;
7 }
-( void ) paintCar: (Car) theCar;
9 @end
11 @implementation PaintShopManager
-( void ) paintCar: (Car) theCar {
13 [subpainter paintCar:theCar];
}
15 @end
1 @interface Painter {}
-(void) paintCar: (Car) theCar;
3 @end
5 @interface PaintShopManager {
Painter subpainter;
7 }
-(void) paintCar: (Car) theCar;
9 @end
11 @implementation PaintShopManager
-(void) paintCar: (Car) theCar {
13 [subpainter paintCar:theCar];
}
15 @end
这样就只剩下左上角标有“聚合”的部分。这是“递归”的另一面，它们共同定义了内聚性的支柱。聚合是将不同的元素组合成一个整体的行为——在本例中，是将看起来与主任务不太相似的子任务（如在“委托”中）组合在一起，但这里只有一个对象完成工作。没有其他对象可以委托或重定向工作，因为这是一个单对象任务，但它被分解为更小的部分，由一个对象内的不同方法处理。这样做可能是为了可读性、灵活性，或者（通常情况下）在定义原子行为之后的某个时间将原子行为组合成一个更复杂的行为。无论如何，它涉及一个对象和不同的方法，如清单 2.10所示，它重温了清单 2.7Timer中的例子。
This leaves just the top left corner, the one marked Conglomeration. This is the flip side of Recursion, and together they define the backbone of cohesion. Conglomeration is the act of bringing together disparate elements into a whole—in this case, pulling together subtasks that look little like the main task (as in Delegation), but here just one object does the work. There are no other objects to delegate or redirect work to because this is a one-object task, but it is broken down into smaller portions that are handled by different methods within the one object. This may be done for readability, flexibility, or, as is often the case, marshaling together atomic behaviors into a more complex one some time after the atomic ones were defined. In any case, it involves one object and disparate methods, as shown in Listing 2.10, which revisits our Timer example from Listing 2.7.
这四个基本概念，递归、委托、重定向和聚合，是四个 EDP。是的，它们很简单。是的，它们是程序员每天都会反射性地使用，但这正是重点，不是吗？这些用法是反射性的，而不是有预谋的，并且每种用法都会对以后的实施和设计问题产生影响。如果您是编程新手，这些概念不是反射性的，尚未根深蒂固，您可能会对它们有疑问。“我什么时候使用它们？”“它们有什么用？”“还有哪些相关概念？”这正是设计模式被创建来表达和传递的那种部落智慧。请提前查看本章后面目录中的 EDP 条目，了解有关这四个 EDP 的更详细讨论。即使您是经验丰富的专业人士，也请这样做 - 那里的内容比您最初想象的要多。
These four basic concepts, Recursion, Delegation, Redirection, and Conglomeration, are four EDPs. Yes, they’re simple. Yes, they’re patterns that programmers use every day by reflex, but that’s rather the point, isn’t it? The uses are reflexive, not premeditated, and each has consequences for later implementation and design issues. If you’re new to programming, these concepts are not reflexive, not yet ingrained, and you may have questions about them. “When do I use them?” “What are they good for?” “What other concepts are related?” This is exactly the type of tribal wisdom that design patterns were created to express and pass along. Look ahead to the EDP entries in the catalog following this chapter for more detailed discussions of these four EDPs. Do so even if you’re a seasoned pro—there’s more there than you may first think.
清单 2.10。Java中Conglomeration方法调用的示例。
Listing 2.10. Example of a Conglomeration method call in Java.
1 class Timer {
void goDing();
3
public void countDown( int counter) {
5 if (counter > 0) {
// 本次忽略此分支
7 } else {
this .goDing();
9 }
};
11 };
1 class Timer {
void goDing();
3
public void countDown(int counter) {
5 if (counter > 0) {
// Ignore this branch this time
7 } else {
this.goDing();
9 }
};
11 };
那么网格中最后两个条目呢？其中对象相似性未知但方法相似性未知？嗯，目前没有人非常确定。如果您能想到这两个空间解决了什么编程概念或设计问题，那么您就有机会添加到 EDP 集合中。毕竟，这并非没有先例。当开发人员偶然发现使用现有部分的新方法时，某些方法将被视为有用并将发展语义。这些新技术如果足够有用，将不可避免地成为下一代语言的主要语言特性。您可能就是帮助定义它们的人。与此同时，让我们删除现有研究所在的中心列和中心行，我们还不确定该怎么处理它们。这给我们留下了一个二乘二网格中的四个条目，如图2.8所示。
What about the final two entries in the grid where the object similarity is unknown but the method similarity is? Well, no one’s quite sure at this point. If you can think of what programming concept or design issue is addressed by those two spaces, there’s an opportunity for you to add to the EDP collection. This isn’t without precedent, after all. As developers stumble on new ways of using established pieces, some ways will be seen as useful and will develop semantics. These new techniques, if useful enough, inevitably end up in the next generation of languages as primary language features. You might be the person who helps define them. In the meantime, let’s remove the center column, where existing research sits, and the center row, which we’re really not sure what to do with yet. This leaves us with four entries in a two-by-two grid, as in Figure 2.8.
图 2.8。我们的前四个 EDP。
Figure 2.8. Our first four EDPs.
有趣的是，我们从编程理论的基本原理中得出了这些简单的概念。一旦我们开始研究对象类型，概念就会变得更加复杂，尽管理论仍然很简单。让我们看看当我们添加第三个轴时会发生什么。
What’s interesting is that we derived these simple concepts from basic principles of programming theory. Once we start playing with the object types as well, the concepts become much more complex even though the theory remains just as simple. Let’s see what happens when we add that third axis.
图 2.9显示了三个轴。之前我们根本不关心对象类型，现在我们可以使用之前为对象类型相似性轴定义的四个条目。未知条目在这里的用处和在上一个网格中一样小，所以让我们再次将其删除。现在我们只关注通过将我们的方法相似性固定为相似性来定义右侧切片，如图2.10所示。该空间的其余部分将在第 5 章中更全面地讨论。
Figure 2.9 shows the three axes. Where before we were not concerned at all with the object type, we can now work with the four entries we defined earlier for the object type similarity axis. The unknown entry has as little utility here, as it did in the previous grid, so let’s again remove it. For now we’ll concentrate on just the right-hand slice defined by fixing our method similarity to similar, as shown in Figure 2.10. The remainder of this space is more fully discussed in Chapter 5.
图 2.9.设计空间扩展到三维。
Figure 2.9. The design space extended to three dimensions.
图 2.10. 方法相似性固定为相似的设计空间。
Figure 2.10. The design space with method similarity fixed to similar.
我们可以通过认识到对象的类型不能完全不同才能使对象相似并调用类似的方法，将递归放在这个网格上。1然而，递归的定义是完全相同的方法实现被其自身调用，这确实需要对象和类型的相似性。
We can place Recursion on this grid by recognizing that the type of the object can’t be completely dissimilar for the object to be similar and calling a similar method.1 The definition of Recursion, however, is that the exact same method implementation is being called by itself, which does require both object and type similarity.
同样，我们可以将“重定向”放在网格中，因为我们知道它最好被描述为一个不同的对象和一个不同的类型。这是“重定向”的最通用形式。但是，如果我们进行增量更改，并且保留对象差异但使对象具有相同的类型，会发生什么？这种常见的设计特征（即一个对象将请求交给相同类型的另一个对象）形成了一条由相似对象组成的链，这些对象在它们之间分配任务。因为这结合了“重定向”（不同的对象）和“递归” （相同类型）的方面，所以我们使用“重定向递归”这个非常巧妙的名称。2
Likewise, we can place Redirection on the grid by realizing that it is best descri bed as a dissimilar object and a dissimilar type. This is the most general form of Redirection. What happens when we make an incremental change, though, and we retain object dissimilarity but make the objects of the same type? This common design trait, where an object hands off a request to another object of the same type, forms a chain of like objects parceling out a task among themselves. Because this combines aspects of both Redirection (a dissimilar object) and Recursion (the same type), we use the extremely clever name of Redirected Recursion.2
让我们继续探索重定向，并稍微改变一下类型关系。如果我们将方法任务重定向到另一个不属于同一类型但属于当前对象的超类型的对象，我们就可以让整个类型家族来处理多态调用。您会注意到，这是我们第一次明确地在讨论中提到多态。到目前为止，我们讨论了相似（相同）类型和不相似（不同）类型之间的调用，但我们还没有细化后一类。现在我们做到了，突然间，面向对象设计的全新层次浮现出来。3我们简单地将这种类型关系称为Subtype。4
Let’s continue this exploration of Redirection and change the type relationship a bit. If we redirect a method task to another object that is not of the same type, but a supertype of the current object, we can let an entire family of types handle the call polymorphically. You’ll notice that this is the first time polymorphism has explicitly popped up in our discussion. Until now, we talked about calls between similar (same) types and dissimilar (different) types, but we hadn’t refined the latter category. Now we have, and suddenly a whole new level of object-oriented design comes to light.3 This type relationship we simply call Subtype.4
我们可以再添加一个不同类型箱的细分，即兄弟类型关系。当所涉及的对象的类型共享一个共同的祖先超类型，但两者都不是对方的超类型时，就会发生兄弟关系。我们将方法调用抛到类型层次结构上，然后将其限制回该树的受信任子集。我们称这种变化为代理重定向。
We can add one more subdivision of the dissimilar type bin, and that is the Sibling type relationship. A sibling relationship occurs when the types of the objects involved share a common ancestor supertype, but neither is a supertype of the other. We tossed the method call up our type hierarchy and then limited it back down to a trusted subset of that tree. We call this variation Deputized Redirection.
让我们回到递归，然后向右滑动一个方格，其中对象实例相同，但类型关系是子类型。再读一遍。谈论拥有相同的对象，以及从该对象到其自身的方法调用，但该方法调用涉及两种对象类型，这实际上是有意义的。事实上，这样做是超级的。
Let’s go back to Recursion, and then slide one square to the right, where the object instance is the same, but the type relationship is a subtype. Read that again. It actually does make sense to talk about having the same object, and a method call from that object to itself, but with two object types involved in that method call. In fact, to do so is super.
具体来说，super可让您从子类型内部访问超类型的方法实现。在 Java 中，这是通过super关键字完成的。在 C++ 中，使用显式类型作用域机制，如Supertype::。其他语言具有多种机制，但它们都做同样的事情 - 它们使对象能够访问超类型的方法实现。这有何用处？好吧，考虑这样一种情况，您想要扩展方法的功能，而不是用新功能完全替换它。您将子类化该类型，重写该方法，然后在您的额外代码中包装对原始方法的调用。正如您所料，这个 EDP 简称为Extend Method。
Specifically, super lets you access a supertype’s implementation of a method from within a subtype. In Java, it is done with the super keyword. In C++, an explicit type-scoping mechanism is used as in Supertype::. Other languages have a variety of mechanisms, but they all do the same thing—they give an object access to a supertype’s implementation of a method. How is this useful? Well, consider a case where you want to extend the functionality of a method, not completely replace it with new functionality. You would subclass the type, override the method, and then wrap a call to the original method in your bit of extra code. As you might expect, this EDP is simply called Extend Method.
请注意，在前面的讨论中，我们采用了一个现有的、熟悉的概念，并在每个阶段将其移开一个空间，每次调整一个细节或另一个细节。这种简单的进展的结果是令人惊讶地收集了编程中的概念和设计元素。如果您不相信我们已经涵盖了很多概念基础，请查看前八个 EDP 的 UML 图的多样性。只考虑涉及递归和代理重定向的想法，如图 2.11和2.12所示。它们看起来一点也不相似，但在单个方法调用的设计空间中，它们仅相距三步。
Notice that in the preceding discussion, we took an existing, familar concept and moved just one space away from it at each stage, tweaking one detail or another at a time. The result of this simple progression is a surprisingly broad collection of concepts and design elements from programming. Look at the variety of the UML diagrams for the preceding eight EDPs if you’re not convinced that we’ve covered a lot of conceptual ground. Consider just the ideas involving Recursion and Deputized Redirection, shown in Figures 2.11 and 2.12. They look nothing alike, yet they sit just three steps from each other in the design space for a single method call.
图 2.11.递归示例 UML。
Figure 2.11. Recursion Example UML.
图2.12.代理重定向示例UML。
Figure 2.12. Deputized Redirection example UML.
同样，还有两个神秘盒子。这次我们不要忽略它们，而是讨论它们在概念上可能意味着什么。在左边的第一个位置，我们有相同（或相似）的对象，但是在方法调用依赖的每一侧类型完全不同。怎么会这样？嗯，我不确定。尽管存在一些相当深奥的类型关系，但这个关系没有多大意义。现在，这还是一个未知领域，但也许将来某种语言会找到这种上下文组合的合理用途。例如， BETA 语言 [ 26 ] 具有一个独特的内部构造，它是的镜像super，从而创建了超类型相似性而不是子类型相似性。然而，这种情况极其罕见，因此未在我们的对象相似性轴中使用。如果它被证明是一个有趣且可行的设计元素，可能会在以后添加它。
Again, two mystery boxes remain. Let’s not ignore them this time but instead discuss what they could mean conceptually. In the first slot on the left, we have the same (or similar) object but utterly dissimilar typing on each side of the method call reliance. How can this be? Well, I’m not sure. Although some rather esoteric typing relationships exist out there, this one doesn’t make a lot of sense. For now, it is unknown territory, but perhaps a language will find a reasonable use for this context combination in the future. The BETA language [26], for example, has a unique inner construct that is the mirror image of super, creating a Supertype similarity instead of a Subtype similarity. This is exceedingly rare, however, and not used in our object similarity axis for that reason. It may be added later if it proves to be an interesting and viable design element.
最右边的插槽也出现了类似的情况。同样，我们有相同（或相似）的对象，但现在是兄弟类型相似性，同样，据我所知，没有一种语言提供这种可能性。这些开放的插槽适合进行解释和思想实验。最重要的是，我们可以预测设计空间中这些未描述的条目的一些属性，就像元素周期表中的元素一样。
An analogous situation occurs for the slot on the far right. Again, we have the same (or similar) object, but now a Sibling type similarity, and again, no language that I am aware of offers this possibility. These open slots are ripe for interpretation and thought experiment. Best of all, we can predict some properties for these undescribed entries in the design space, much like elements in a periodic table.
这仅仅是方法调用 EDP 的一半多一点，稍后将在参考部分完整介绍这些方法调用。一次方法调用就能表达这么多概念，真是令人惊叹，不是吗？一次方法调用。这就是我们讨论的全部内容，但我们还有各种各样的设计组件可供使用。
That’s just over half the method call EDPs that will be presented in full later in the reference section. It’s remarkable how many concepts can be expressed with a single method call, isn’t it? One method call. That’s all we’ve been discussing, and yet we have a wide range of design components to play with.
2.3. 核心 EDP
2.3. Core EDPs
到目前为止，我们只讨论了方法调用，但回想一下，我们说过，我们可以使用其他三种形式的依赖：字段使用、状态更改和凝聚力。这些呢？嗯，目前它们仍在探索、充实和撰写中。首先讨论方法调用，因为它们是最小、最简单的设计空间。其他的要大得多。然而，为了填补第一组，应该解决一些基本概念，它们从形式和实用的角度构成了面向对象编程的核心。
To this point, we’ve discussed just method calls, but recall that we said there are three other forms of reliances that we can play with: field usage, state changes, and cohesion. What about those? Well, they are still being explored at the moment, fleshed out, and written up. Method calls were approached first because they’re the smallest, simplest design space to work with. The others are quite a bit larger. There are a few basic concepts that should be addressed, however, to fill out this first group, and they form the core of object-oriented programming from both formal and pragmatic points of view.
首先，有整个“创建对象”的想法。毕竟，这是面向对象编程的显著特征之一，否则整个范式可能会被命名为其他名称。请参阅此处的Create Object EDP 以获取指导。从这个角度来看，面向对象编程已经发展了三十年左右，人们竟然反对它作为一种编程方法，这似乎很奇怪，但它有很大的障碍需要克服。一个反对意见是，任何可以在面向对象编程中实现的东西都可以在过程编程中实现。虽然从技术上讲这是正确的——任何在高级系统中可行的事情都可以在过程、汇编程序或原始二进制中实现；否则它就无法运行——但面向对象编程允许轻松执行某些原则。Create Object概述了其中一个主要原则，即同时在单个对象和单个类型之间建立关系。
To begin with, there’s the entire “making objects” idea. It is one of the distinguishing characteristics of object-oriented programming, after all, or the entire paradigm probably would have been named something else. Look at the Create Object EDP for guidance here. At this vantage point, three decades or so into object-oriented programming, it seems odd that there was ever objection to it as a programming approach, but it had large hurdles to overcome. One objection was that anything that can be implemented in object-oriented programming can be implemented in procedural programming. While technically true—anything doable in any higher-level system is doable in procedural, in assembler, or in raw binary; otherwise it wouldn’t run—object-oriented programming allowed the easy enforcement of certain principles. Create Object outlines one of the major ones, creating a relationship between a single object and a single type at the same time.
一旦我们创建了对象，我们如何让它们相互通信？显然，通过使用方法调用。好吧，但是我们首先如何让它们找到彼此？我们可以静态地创建对象之间的所有连接，但这样做相当有限，因为这意味着系统无法对运行时输入做出反应。程序员必须预测系统的所有可能用途，包括所有可能的数据需求的最大内存使用量等细节，而不管实际使用量是多少。为了解决这个问题，我们必须允许对象在执行期间访问其他对象。Retrieve EDP 可以做到这一点，动态地在对象之间创建运行时连接。
Once we create objects, how do we get them talking to one another? Well, by using method calls, obviously. Okay, but how do we get them to find one another in the first place? We could statically create all the connections between the objects, but that’s rather limiting because it means the system cannot react to runtime input. The programmer must anticipate all possible uses of the system, including details such as maximal memory use for all possible data needs, regardless of the actual amount used. To get around this problem, we have to allow objects to gain access to other objects during execution. The Retrieve EDP does this, dynamically creating runtime connections between objects.
要实例化一个对象，我们需要一个类型来实例化。通常，我们可用的类型几乎是正确的，但并不完全正确。不过，我们不想每次都从头开始重写一个合适的类型，而是想利用已有的类型作为起点。重用类型的能力是通过另一个核心 EDP继承实现的。继承是一种类型依赖，我们之前提到过它的一个别名，子类型。它在基类型和依赖它实现大部分基本功能（方法）和状态（字段）的类型之间建立了联系。它允许新类型以格式良好的方式重用大量逻辑和数据。
To instantiate an object, we need a type from which to do so. Often the types available to us are almost, but not quite, correct. Instead of rewriting a proper type from scratch every time, though, we’d like to leverage what we already have as a starting point. The ability to reuse types is made possible through Inheritance, another core EDP. Inheritance is a type reliance, which we mentioned earlier by one of its aliases, subtyping. It forms a connection between a base type and a type that relies on it for much of its basic functionality (methods) and state (fields). It allows a new type to reuse large chunks of logic and data in a well-formed way.
另一种与类型相关的 EDP 是抽象接口。它形成两种类型之间的依赖关系，但略有不同，因为关系的一端是预先知道的。它使用方法作为两种类型之间的桥梁。如果类型将方法声明为抽象的，则不必这样做——严格来说，这是不允许的 给出方法主体实现。此 EDP 创建一个承诺，即在某个时候另一个类型将从此类型继承并为该方法提供适当的实现。该类型是什么尚不清楚，但必须发生这种情况才能使此类型可用。在此之前，此类型是不完整的。
Another type-related EDP is Abstract Interface. It forms a reliance between two types that is a bit different in that only one end of the relationship is known ahead of time. It uses a method as a bridge between the two types. If a type declares a method to be abstract, it doesn’t have to—and in the strictest sense isn’t allowed to—give a method body implementation. This EDP creates a promise that at some point another type will inherit from this one and give the method a proper implementation. What that type is isn’t yet known, but it has to happen for this type to be usable. Until then, this type is incomplete.
这四个 EDP（创建对象、检索、继承和抽象接口）允许我们创建强制执行某些保证的对象、在运行时将它们相互关联、根据其他类型定义对象类型以及为未来未指定的类型声明承诺。它们共同构成了面向对象编程的大部分基础。与方法调用 EDP（如委托和递归）一起，它们为将设计视为可重复的学科提供了一个坚实的开端，该学科使用以有条不紊且精确的方式连接起来的格式良好的构建块。为什么我们不看看我们能用其中的一小部分做些什么呢？
These four EDPs, Create Object, Retrieve, Inheritance, and Abstract Interface, allow us to create objects with enforcement of certain guarantees, relate them to one another at runtime, define object types in terms of other types, and declare promises for future, unspecified types. Together, they form much of the basis for object-oriented programming. Along with the method call EDPs such as Delegation and Recursion, they provide a solid start to treating design as a reproducible discipline, one that uses well-formed building blocks connected in methodical and precise ways. Why don’t we see what we can do with just a handful of them?
2.4. 结论
2.4. Conclusion
本章向您介绍了基本设计模式。它简要概述了它们帮助解决的主要问题以及它们的产生背景。您看到了如何用较小的模式来描述诸如Decorator之类的模式。这导致得出这样的结论：为了更好地描述标准文献中更抽象的设计模式，更细粒度的模式是必要的。
This chapter introduced you to the Elemental Design Patterns. It gave a brief synopsis of the driving problem they helped solve and some background on how they came about. You saw how patterns such as Decorator can be described in terms of smaller patterns. This led to the conclusion that to better describe the more abstract design patterns in the standard literature, finer-grained patterns are necessary.
我们向您介绍了一种最简化的面向对象编程理论，并向您展示了它如何产生编程中可能出现的少量关系或依赖关系。这些依赖关系构成了我们可以定义的最小模式的基础。其中一种依赖关系，即方法调用依赖关系，构成了本书的核心。您看到每个方法调用都有一个上下文，由三个简单的信息（对象、类型和方法相似性）定义，并且这些信息创建了一个设计空间，其中存在众所周知的编程概念。其他三个依赖关系（字段使用、状态更改和内聚性）产生了它们自己的设计空间，这些设计空间正在被定义。
You were introduced to a minimalist form of object-oriented programming theory and shown how it gives rise to a small number of possible relationships, or reliances, that can occur in programming. These reliances form the basis for the smallest patterns we can define. One of these reliances, the method-call reliance, forms the core of this book. You saw that every method call has a context, defined by three simple pieces of information—the object, type, and method similarities—and that these create a design space in which well-known programming concepts live. The other three reliances—field use, state change, and cohesion—give rise to their own design spaces that are in the process of being defined.
简单探索了方法调用设计空间。为许多 EDP 提供了代码示例，并让您了解了它们之间的关系。最后，您学习了一组核心 EDP，它们定义了面向对象编程的许多基本概念。
The method-call design space was explored lightly. Code examples were given for many of the EDPs, and you saw how they relate to one another. Finally, you learned a set of the core EDPs, which define many of the basic concepts that underlie object-oriented programming.
现在，您已经具备了坚实的基础来了解 EDP 的重要性和实用性，并准备开始学习如何使用它们。在第 3 章中，您将学习一种有价值的方法，以图形方式描述模式实例，以帮助您直观地了解它们的交互。
You now have a solid footing to understand the importance and utility of the EDPs and are ready to start learning how to work with them. In Chapter 3, you’ll learn a valuable way to graphically depict pattern instances to help you visualize their interactions.
3. 模式实例符号
3. Pattern Instance Notation
在我们探讨元素设计模式 (EDP) 的影响之前，我想先稍微岔开话题，介绍一种新的图形符号，即模式实例符号 (PIN)。从现在开始，PIN 将帮助您直观地了解我们讨论的一些概念。
Before we explore the impact of what we can do with Elemental Design Patterns (EDPs), I want to take a bit of a side jaunt and introduce a new graphical notation, the Pattern Instance Notation, or PIN. PIN is used from here out to help you visualize some of the concepts we discuss.
本章非正式地描述了 PIN 及其在本书中的使用方式。如果您有兴趣了解更多信息或了解如何将 PIN 用于软件设计的工具支持，请参阅“模式实例符号：用于动态可视化和理解软件模式的简单分层可视化符号”[ 36 ] 中对 PIN 的完整描述。
This chapter provides an informal description of PIN and how it is used in this book. If you’re interested in further information or in how PIN can be used in tool support for software design, PIN is fully described in “The Pattern Instance Notation: A Simple Hierarchical Visual Notation for the Dynamic Visualization and Comprehension of Software Patterns” [36].
3.1. 基础知识
3.1. Basics
PIN 是构成模式的概念和想法的视觉表示。它旨在作为一种快速简便的方式来记录和描述设计模式及其交互。之所以选择 PIN 这个名称，是因为它在以其他图表符号（例如 UML）显示模式或概念的实例时也很有用。这是最初的用例。
PIN is a visual representation of the concepts and ideas that comprise patterns. It is intended as a quick and easy way to document and describe design patterns and their interactions. The name PIN was chosen because it also is useful in showing instances of patterns or concepts in other diagramming notations, such as UML. This was the original use case.
在第 2 章2.2.2节中，我提到了设计模式规范的参与者部分。参与者是设计模式的一部分，设计模式实例必须存在。它们的另一个名称是设计模式的角色，因为每个参与者在整合模式方面都扮演着特定的角色。换句话说，每个参与者都履行模式的一个角色。当我们说“装饰器模式中名为ConcreteDecorator的参与者”时，我们真正的意思是“履行装饰器模式中 ConcreteDecorator 角色的实现类”。角色也是抽象的，因为它们为将充当参与者的实现或设计特性提供了名称和约束。
In Chapter 2, Section 2.2.2, I mentioned the Participants section of a design pattern specification. Participants are the parts of a design pattern that must exist for an instance of the design pattern to occur. Another name for them are the roles of the design pattern, because each participant has a particular role to play in bringing together the pattern. In other words, each participant fulfills a role of the pattern. When we say “the participant named ConcreteDecorator of the Decorator pattern,” what we really mean is “the class of the implementation that fulfills the ConcreteDecorator role of the Decorator pattern.” The roles are abstractions as well, because they provide a name for, and constraints on, the implementation or design features that will act as participants.
它与戏剧（例如《哈姆雷特》）没有什么不同。哈姆雷特是一个角色，奥菲莉亚、罗森克兰兹和盖尔登斯顿也是角色。它们描述了一个角色，并为扮演每个角色的演员类型提供了指导。演员是舞台制作的参与者，所有演员共同合作形成它。
It’s not unlike a play, such as Hamlet. Hamlet is a role, as are Ophelia, Rosen-crantz, and Guildenstern. They describe a part and provide guidelines for the kind of actor who will be cast in each role. The actor is a participant in the stage production, and all the actors collaborate to form it.
戏剧类似于设计模式，而特定的舞台作品则是该模式的一个实例。剧本就像设计模式规范：它描述了角色，并描述了他们如何交流和互动。扮演这些角色的演员就是参与者。
The play is analogous to a design pattern, and the particular stage production is an instance of the pattern. The script is like the design pattern specification: it describes roles, and it describes how they communicate and interact. The actors who are cast in those roles are the participants.
UML 有一个图形特性，即协作元素，可以描述这种情况。图 3.1显示了一个使用装饰器的示例。这绝不是一个糟糕的符号，它非常灵活，但是对于我们的需求来说，它存在一些问题。首先，除了在简单的图表中，它在实际使用中会有些混乱。它只会向 UML 图添加信息。使用得越多，图就越复杂。设计模式等抽象旨在通过让我们在更高的抽象层次上操作来减少复杂性和细节。UML 协作的症结在于它们无法降低 UML 图的复杂性。它们不能像更高的抽象层次那样减少暴露的信息量。首先，协作元素还需要放入完整的 UML 图中。当我们讨论 UML 特性时，这似乎是一个奇怪的抱怨，但事实证明，在很多情况下，我们可能希望讨论模式如何交互，而不必为实现它们所需的所有部分创建图表。
UML has a graphical feature, the collaboration element, to describe such situations. An example using Decorator is shown in Figure 3.1. This isn’t a bad notation by any means, and it is very flexible, but it suffers from a couple of problems for our needs. For one thing, it is a bit messy in practical use at anything other than in trivial diagrams. It only adds information to a UML diagram. The more you use it, the more complex the diagram becomes. Abstractions such as design patterns are intended to reduce complexity and detail by letting us operate at a higher level of abstraction. A sticking point of UML collaborations is that they aren’t capable of reducing the complexity in a UML diagram. They can’t reduce the amount of exposed information as a higher level of abstraction should. A collaboration element also requires a complete UML diagram to be placed in, in the first place. This may seem like an odd complaint when we’re discussing a UML feature, but it turns out that there are plenty of situations where we may want to discuss how patterns interact without having to create a diagram for all the pieces necessary to implement them.
图 3.1. UML 协作图。
Figure 3.1. UML collaboration diagram.
设计模式的另一种常见符号是四人帮 (GoF) [ 21 ] 引入的模式：角色符号。这也是一种在 UML 上使用的注释，它使用以模式和角色命名的外部标志。此注释可以根据需要应用于类、包、字段或方法，如图3.2所示。它非常灵活且显而易见，但并非没有问题。
Another common notation for design patterns is the pattern:role notation introduced by the Gang of Four (GoF) [21]. It is also an annotation to be used on top of UML, and it uses an external flag that is named with the pattern and the role. This annotation can be applied to a class, package, field, or method as needed, as shown in Figure 3.2. It’s very flexible and obvious, but it isn’t without issues.
图 3.2. UML 中的策略模式：角色标签。
Figure 3.2. Strategy as pattern:role tags in UML.
当只涉及并显示一个图案时，这种方法还不错，但是当你将一个图案隐藏在包含数百甚至数千个碎片的图表中时会发生什么？比如说，像图 3.3中那样？
This approach isn’t so bad when only one pattern is involved and displayed, but what happens when you have a pattern hidden in a diagram with hundreds or even thousands of pieces? Say, something like in Figure 3.3?
图 3.3. 一个不太大的系统的大型 UML。
Figure 3.3. Huge UML of a not-so-huge system.
图 3.3是来自实际项目的实际 UML 图。在这种比例下它是不可读的，但实际上，在任何比例下它几乎都是不可读的。如果以实际尺寸打印它，例如字体为 10 点大小，它将是 86 x 32英尺。这需要很多纸张：3,290 张标准美国信纸大小的纸张。在这种尺寸下，它太小而无法阅读，但在完整尺寸下，它太大而无法进行视觉搜索。令人沮丧的是，图表中记录了大量模式，可供查找和学习，以帮助记录系统。
Figure 3.3 is an actual UML diagram from an actual project. It’s unreadable at this scale, but practically speaking, it’s almost unreadable at any scale. If you were to print it at actual dimensions, such that the type was 10-point size, it would be 86 by 32 feet. That’s a lot of paper: 3,290 standard U.S. letter-size sheets. At this size, it’s too small to be read, but at the full size, it’s too large to be searched visually. Frustratingly, a wealth of patterns are recorded in the diagram, ready to be found and learned from to help document the system.
举一个更具体的例子，考虑如图 3.4所示的 UML 图。图 3.4中有两个策略模式实例，但使用模式：角色符号，即使在这个小例子中，我们也无法分辨哪些部分与哪个模式实例相匹配。
As a more concrete example, consider a UML diagram such as in Figure 3.4. There are two instances of the Strategy pattern in Figure 3.4, but using the pattern:role notation, we can’t tell which pieces go with which pattern instance, even in this small example.
图 3.4. UML 中作为模式：角色标记的策略的多个实例。
Figure 3.4. Multiple instances of Strategy as pattern:role tags in UML.
这就是 PIN 的最初目的：能够将单个图表中的多个模式实例清楚地表达为一级实体。因此得名“模式实例符号”。其他目标是使符号简单、灵活，并且可以与其他符号（如 UML）结合使用，而无需依赖它们。
This was the initial reason for the PIN: to be able to clearly express multiple instances of patterns in a single diagram as first-class entities. Hence the name, Pattern Instance Notation. Other goals are that the notation be simple, flexible, and usable in conjunction with other notations such as UML without being reliant on them.
3.2. PIN 盒
3.2. The PINbox
PIN 中的核心符号是 PINbox。它代表单个模式实例，并允许您选择要公开的粒度级别。我们从最简单的 PINbox 开始，然后逐步研究更有趣的变体。
The core notation in PIN is the PINbox. It represents a single pattern instance and lets you choose the level of granularity you want to expose. We start with a PINbox in its simplest form and work our way up into more interesting variants.
3.2.1. 折叠 PIN 框
3.2.1. Collapsed PINbox
折叠的 PINbox 只是一个带有粗双边框的名称标签，如图3.5所示。内部边框笔触为矩形，外部边框笔触具有弧形角。如果可能，边框会用灰色阴影显示。框上的标签是模式的名称。
The collapsed PINbox is just a name label with a thick double border, as in Figure 3.5. The interior border stroke is a rectangle, the external border stroke has curved corners. If possible, the border is shaded gray. The label on the box is the name of the pattern.
图 3.5. 折叠的 PINbox。
Figure 3.5. Collapsed PINbox.
就是这样。这是 PINbox 的最简单形式。别担心，后面还会有更多内容。不过，即使在这里，我们也有一些有用的东西可以使用。我们可以将其用作 UML 或其他图表上的注释，只需从 PINbox 向图表中与其最密切相关的任何元素绘制一个箭头即可，如图3.6所示。当谈到图形符号时，经验表明，框和线是最简单的。
That’s it. That’s the simplest form of a PINbox. Don’t worry, there’s more to come. Even here, though, we have something useful to work with. We can use this as an annotation onto a UML or other diagram by just drawing an arrow from the PINbox to whatever element in the diagram is most closely associated with it, as in Figure 3.6. When it comes to graphical notations, experience shows that boxes and lines are about as simple as it gets.
图 3.6.折叠的 PINbox 作为注释。
Figure 3.6. Collapsed PINbox as annotation.
有关这在现实生活中如何运作的一些示例，请参阅图 3.7和3.8 。使用Singleton和Abstract Factory 的示例显示了 PIN 与类图一起使用，而Template Method表示序列图中的模式。
For some examples of how this might work in real life, consider Figures 3.7 and 3.8. The example using Singleton and Abstract Factory shows PIN being used with a class diagram, while Template Method indicates a pattern in a sequence diagram.
图 3.7.类图中的单例和抽象工厂。
Figure 3.7. Singleton and Abstract Factory in class diagram.
图 3.8.序列图中的模板方法。
Figure 3.8. Template Method in sequence diagram.
折叠式 PINbox 最适合于您需要助记符、快速提醒系统中存在某种模式，并且该模式具有您可以指出的单一定义特征的情况。这是一种非正式的符号，但很有用，尤其是在勾勒新设计时。
The collapsed PINbox is best used when you need a mnemonic, a quick reminder that a pattern exists in a system, and the pattern is one that has a single, defining feature you can point to. This is an informal notation but one that has utility, particularly when sketching out a new design.
3.2.2. 标准 PIN 盒
3.2.2. Standard PINbox
更上一层楼，我们有了标准的 PINbox。我们扩展了折叠表单的粗灰色边框，以便可以向其中添加文本。具体来说，我们添加组成中间列出的模式的角色，如图3.9所示。角色的命名使用第 2 章第2.2.1 节中概述的相同约定。添加角色让我们能够表达一些明显更细粒度的连接。例如，由于每个模式都有一组格式良好且不同的角色，我们可以将每个角色连接到 UML 类或序列图的元素，如图3.10和3.11所示。
Taking things up a level, we have the standard PINbox. We expand the thick gray border of the collapsed form so that we can add text to it. Specifically, we add the roles that make up the pattern listed in the middle, as in Figure 3.9. The roles are named using the same conventions outlined in Chapter 2, Section 2.2.1. Adding the roles lets us express some significantly finer-grained connections. For instance, because each pattern has a well-formed and distinct set of these roles, we can connect each role to the elements of a UML class or sequence diagram, as in Figures 3.10 and 3.11.
图 3.9. 标准 PINbox。
Figure 3.9. Standard PINbox.
图 3.10. PIN 与 UML 类图一起使用。
Figure 3.10. PIN used with UML class diagram.
图 3.11. PIN 与 UML 序列图一起使用。
Figure 3.11. PIN used with UML sequence diagram.
边框周围角色名称的排列由您决定；但只要它在最终图表上看起来最好就可以了。任何顺序都可以，只要能让最终图表最容易阅读即可。
The arrangement of the role names around the border is up to you; however it looks best on the final diagram is fine. Whatever ordering makes the final diagram easiest to read is preferred.
这看起来非常像图 3.1中的 UML 协作，但 PINbox 可以做一些协作无法做到的事情。请注意，在图 3.11中，享元实例的AbsFlyweight 角色未附加到任何内容。在这种情况下，这很好，因为我们只是试图描述其他三个角色之间的交互。如果我们尝试使用协作元素并希望保留有关享元模式角色的所有信息，我们要么必须使用未使用的角色名称的悬垂箭头，要么更糟的是，根本不显示它。两者都不是最佳选择。使用 PINbox，我们不会丢失信息，也不会不必要地弄乱我们的图表。再举一个例子，假设我们在软件设计中有两个模式实例，我们想要指出在一个模式中履行角色的同一个类在另一个模式中也履行角色，将这两个模式实例绑定到一个组合结构中。我们可以像图 3.12中那样做。请注意，此图未使用任何 UML；它仅以特定方式显示两种模式之间的联系。当我们将多个 PINbox 连接在一起时，事情会变得更加有趣。
This looks an awful lot like a UML collaboration from Figure 3.1, but PINboxes can do some things that collaborations can’t. Note that in Figure 3.11, the instance of Flyweight has the AbsFlyweight role not attached to anything. This is fine in this case because we’re just trying to depict the interactions between the other three roles. If we tried this with a collaboration element and wanted to retain all of the information about the roles of the Flyweight pattern, we’d either have to have a dangling arrow with the unused role name or, worse, not show it at all. Neither is optimal. With the PINbox, we don’t lose the information, and we don’t clutter up our diagram needlessly. For another example, say we have two pattern instances in a software design and we want to indicate that the same class that fulfills a role in one fulfills a role in another, tying the two pattern instances into a combined structure. We could do this as in Figure 3.12. Notice that this diagram doesn’t use any UML; it only shows the connection between two patterns in a particular way. Things get much more interesting when we connect multiple PINboxes together.
图 3.12. 标准 PIN 角色连接。
Figure 3.12. Standard PIN role connections.
没有 UML。
There is no UML.
没有代码。
There is no code.
没有类、方法、字段，什么都没有，只有两个模式、两个概念之间的关系，这种关系以特定的方式将它们联系在一起方式。我们将使用此功能在下一章中展示如何将模式组合成新概念。
There are no classes, methods, fields, no anything, just a relationship between two patterns, two concepts, and that relationship ties them together in a particular way. We use this capability to show how patterns can be combined into new concepts in the next chapter.
但我们还可以做得更多。
We can do still more, though.
3.2.3. 扩展 PIN 框
3.2.3. Expanded PINbox
如果我们对 PINbox 稍加调整，它就会变得更加灵活和强大。让我们从扩展中心的框开始，也就是当前仅包含模式名称的框。我们最终会得到如图 3.13所示的效果。现在我们可以使用这个新的空白区域来绘制更多 PINbox。为什么要这样做？请看图 3.14。这里我们已填充画布。这表明模式具有五个角色，并且可以分解为两个较小的子模式：子模式 A 和子模式 B。此外，它还准确地显示了外环上的角色如何映射到内部子模式的角色。如果我们有子模式 A 和子模式 B，并且子模式 A 中的角色 3 由系统中与子模式 B 中的角色 1 相同的实体履行，那么我们可以用模式的单个实例替换这两个实例。换句话说，我们可以稍微提高抽象级别。我们不需要跟踪两件事，而只需要管理一个模式实例。
If we make a small tweak to the PINbox, it becomes much more flexible and powerful. Let’s start by expanding the box in the center, the one that currently just holds the pattern name. We end up with something that looks like Figure 3.13. Now we can use this new blank area to draw additional PINboxes. Why would we want to do this? Well, look at Figure 3.14. Here we’ve filled in the canvas. What this shows is that Pattern has five roles and can be decomposed into two smaller subpatterns: Subpattern A and Subpattern B. Further, it shows exactly how the roles on the outside ring map to the roles of the inner subpatterns. If we have Subpattern A and Subpattern B, such that Role 3 in Subpattern A is fulfilled by the same entity in a system as Role 1 from Subpattern B, then we could replace those two instances with a single instance of Pattern. In other words, we could raise the level of abstraction a bit. Instead of having to keep track of two things, we only have to manage one pattern instance.
图 3.13.空白扩展 PIN 实例。
Figure 3.13. Blank expanded PIN instance.
图 3.14. 扩展的 PIN 实例。
Figure 3.14. Expanded PIN instance.
另一方面，如果我们有一个Pattern实例，那么显然我们也有每个子模式的实例。它们可能被埋没了一点，但我们知道它们必须在那里。
On the other hand, if we had an instance of Pattern, it’s clear that we also have an instance of each of the subpatterns. They may be buried a bit, but we know they have to be there.
扩展的 PINbox 让我们能够以任意程度揭示模式的层次性质。我们可以将单个模式保留在最高抽象级别，也可以继续扩展它以显示更多细节。此外，如图3.15所示，我们不仅限于在此画布上仅使用 PIN。我们可以集成 UML 或其他符号来说明某一点。对于无法分解为更小的子模式的 EDP，显示 UML 的能力可以提供直接定义。或者，我们可能希望将演示装饰器模式的一种实现的 UML 图放置在 PINbox 内。这可以用作学生的参考图或设计支持工具中的快速提醒。灵活性使 PIN 在描述模式的定义和组成方面如此有用。
The expanded PINbox lets us reveal the hierarchical nature of patterns to whatever degree we wish. We can leave a single pattern at the highest level of abstraction, or we can keep expanding it to show more detail. Also, as shown in Figure 3.15, we are not limited to using just the PIN on this canvas. We can integrate UML or some other notation to illustrate a particular point. In the case of EDPs, which cannot be decomposed into smaller subpatterns, the ability to display UML offers a direct definition instead. Alternatively, we may want to place the UML diagram that demonstrates one implementation of, say, the Decorator pattern, inside the PINbox. This could be used as a reference illustration for a student or as a quick reminder in a design support tool. Flexibility is what makes PIN so uniquely useful for depicting the definition and composition of patterns.
图 3.15. 使用 UML 扩展的 PIN 实例。
Figure 3.15. Expanded PIN instance using UML.
3.2.4. 堆叠 PINbox 和多重性
3.2.4. Stacked PINboxes and Multiplicity
许多模式的元素都具有多重性。例如，通常，在装饰器模式中，不止一个参与者会扮演具体装饰器的角色，而如果只有一种工厂，抽象工厂就会失去其大部分用处。使用对 PINbox 定义的严格解释，我们必须使用一个为每个元素组合创建一个 PINbox。查看图 3.16中的基本抽象工厂图。为了用标准 PINbox 正确注释它，我们需要四个 PINbox：每个具体工厂乘以每个具体产品一个。我们现在只讨论两个工厂和两个产品。想象一下，如果有四个工厂和十种产品。四十个 PINbox 将无法导航且极其混乱。我们会这样做，但我们也希望确保查看该图的人知道这些单独的模式实例只是更高级别的同一模式实例的一部分，它们从根本上是相连的。我们可以使用堆叠 PINbox 来显示关系，如图3.17所示。
Many patterns have a multiplicity to their elements. Generally, more than one participant fulfills the Concrete Decorator role in a Decorator pattern, for instance, and Abstract Factory loses much of its usefulness if there is only one kind of factory. Using a strict interpretation of the PINbox definition, we’d have to use one PINbox for each combination of the elements. Look at the basic Abstract Factory diagram in Figure 3.16. To properly annotate it with standard PINboxes, we’d need four: one for each Concrete Factory times each Concrete Product. And we’re only discussing two factories and two products right now. Imagine if there were, say, four factories and 10 products. Forty PINboxes would be impossible to navigate and horrendously confusing. And we would but we also want to make sure that someone looking at the diagram knows that these individual pattern instances are simply parts of the same pattern instance at a higher level, that they are fundamentally connected. We can show the relationships by using a stacked PINbox, as in Figure 3.17.
图 3.16. 需要多个相关的 PINbox。
Figure 3.16. A need for multiple related PINboxes.
图 3.17. 堆叠 PINbox。
Figure 3.17. Stacked PINbox.
我们在 PINbox 的“后面”添加了第二个轮廓，使 PINbox 看起来像一副纸牌一样堆叠在一起。另外，请注意，我们更改了来自具体工厂、具体产品和抽象产品角色的箭头。现在，尾部呈分叉状，表示存在多个连接，然后线条分叉，指向之前由唯一 PINbox 指示的每个实体。为了帮助说明分叉发生的位置，在连接点处添加了一个小圆圈。缺少此指示器的连接点是多个角色由同一元素满足的位置，与以前一样。
We added a second outline “behind” the PINbox to give the appearance of PINboxes stacked like a deck of cards. Also, notice that we changed the arrows coming from the Concrete Factory, Concrete Product, and Abstract Product roles. The tail now has a forked appearance to indicate that there are multiple connections, and then the line splits to point to each of the entities that previously would have been indicated by a unique PINbox. To help illustrate where the splits occur, a small circle is added at the junction points. Junction points that lack this indicator are where multiple roles are satisfied by the same element, as before.
一般而言，应谨慎使用 PINbox 的多重性连接。请记住，我们的目标是简单，如果两个角色具有与之相关的多重性，我们必须非常清楚，这种关系是由我们表示的抽象正确定义的。在这种情况下，拥有三个具有多重性的角色是有意义的。每个具体工厂都必须处理每个具体产品，而抽象产品显然与其相应的具体子类相关联。在这种情况下，很难混淆。
In general, multiplicity connections should be used with care with PINboxes. Remember our goal is simplicity, and if two roles have multiplicities associated with them, we have to be extremely clear that the relationship is properly defined by the abstraction we are denoting. In this case, having three roles with multiplicity makes sense. Each Concrete Factory must handle each of the Concrete Products, and the Abstract Products are clearly linked with their corresponding concrete subclasses. It’s hard to get confused in this case.
但是，如果有多个抽象工厂实体，则每个实体都需要创建一个新的堆叠 PINbox 实例。这样想：如果一个模式非常适合用作折叠 PINbox，那么当且仅当使其在折叠形式中有用的角色与多重性无关时，它才可能用作具有多个多重性的堆叠 PINbox 。换句话说，如果有一个角色非常突出，以至于它可以在折叠形式中成功使用，那么它可以用作堆叠 PINbox 的关键角色。
If, however, there were multiple Abstract Factory entities, each would necessitate the creation of a new stacked PINbox instance. Think of it this way: if a pattern lends itself well to use as a collapsed PINbox, then it likely will be useful as a stacked PINbox with several multiplicities, if and only if the role that makes it useful in the collapsed form is not associated in a multiplicity. In other words, if there is one role that is so prominent that it can be used in the collapsed form successfully, then it can be used as the linchpin role for a stacked PINbox.
在此基础上，图 3.18重新审视了图 3.4中的情况，当时我们有多个策略实例，但无法可靠地区分它们。现在我们有两个策略集群，每个集群都是不同的，可以清晰区分。我们使用堆叠形式来表示每个模式实例有多个具体策略。这当然很有用，但如果我们想捕捉我们正在协同使用多个策略集群的想法，该怎么办？——而不是单个策略，但不止一个，共同作用。换句话说，如果我们认为这种多策略情况很常见，以至于我们想清楚、清晰地告诉别人，会怎么样呢？在这种情况下，我们可以使用图 3.19来解释它。我们剥离了 UML，即与我们的实现最密切对应的部分，并将策略集群之间的连接抽象为简单而精确的图。该图比图 3.4中的原始图简单得多，并提供了一种更快共享和更清晰地传达我们试图传达的内容的机制：“这里有不止一个策略在起作用，相同的类充当每个策略的具体策略。”这甚至可以成为新设计模式写作的种子。
Building on this, Figure 3.18 revisits the situation from Figure 3.4, where we had multiple instances of Strategy but couldn’t reliably distinguish between them. Now we have two Strategy clusters, and each is distinct and clearly discernable. We use the stacked form to indicate that there are multiple concrete strategies for each pattern instance. This is useful, certainly, but what if we wanted to capture the very idea that we are using multiple clusters of Strategy in concert?—not a single Strategy, but more than one, acting together. In other words, what if we decided that this multiple Strategy situation was common enough that we wanted to tell someone else about it, clearly and cleanly. In such a case, we might use Figure 3.19 to explain it. We stripped out the UML, the portion that corresponds most closely to our implementation, and we abstracted out the connection between the Strategy clusters into something simple and precise. This diagram is much simpler than our original in Figure 3.4 and provides a mechanism for faster sharing and clearer communication of what we are trying to get across: “There is more than one Strategy at work here, and the same classes are acting as concrete strategies for each.” This could even be the seed of a new design pattern writeup.
图 3.18.多个策略实例作为 PINbox。
Figure 3.18. Multiple Strategy instances as PINboxes.
图 3.19.显示多个策略PINbox之间的交互。
Figure 3.19. Showing the interaction between multiple Strategy PINboxes.
3.2.5. 剥离和聚结
3.2.5. Peeling and Coalescing
扩展 PINbox 的最后一项技巧很有用，尽管我们在本书中没有广泛使用它。回想一下，在讨论 UML 协作符号时，我提到它只能在图表中添加信息，而不能减去信息，并且由于这一限制，它无法用作大规模的抽象机制。
There’s one last trick of the expanded PINbox that is useful to know about, even though we don’t use it extensively in this book. Recall that in the discussion of the UML collaboration notation, I mentioned that it can only add, not subtract, information in a diagram and that it fails to be useful as an abstraction mechanism at large scales because of that limitation.
想象一下，我们正在查看如图 3.17所示的系统。我们已经准确地识别了抽象工厂设计模式的多个实例的堆叠 PINbox ，但我们没有使图表变得更简单。这只是适度有用。
Imagine that we’re looking at a system such as in Figure 3.17. We’ve accurately identified the stacked PINbox for the multiple instances of the Abstract Factory design pattern, but we haven’t made the diagram any simpler. This is only moderately useful.
想象一下，图 3.17中的 UML 片段是一个更大图表的一部分。让我们向它添加一些连接但让它们保持模糊，如图3.20所示。很难看出添加了什么，不是吗？现在，让我们将图 3.17中的 UML 部分放置 在PINbox 的展开形式内，如图3.21所示。相同的信息仍然存在，但我们现在使用 PINbox 作为代理机制。附加到角色外边缘的任何东西都会连接到附加到角色内边缘的东西。您可以将灰色角色边框视为直通层。
Imagine that the UML snippet in Figure 3.17 is instead part of a much larger diagram. Let’s add some connections to it but leave them ambiguous, as in Figure 3.20. It’s difficult to see what was added, isn’t it? Now, let’s place the UML portion from Figure 3.17 inside an expanded form of the PINbox, as in Figure 3.21. The same information is still in place, but we’re now using the PINbox as a proxy mechanism. Anything attached to the outside edge of a role connects to what is attached to the inside edge of a role. You can think of the gray role border as a pass-through layer.
图 3.20.抽象工厂作为更大的 UML 图的一部分。
Figure 3.20. Abstract Factory as part of a larger UML diagram.
图 3.21.扩展的 PINbox 中包含抽象工厂。
Figure 3.21. Abstract Factory subsumed within the expanded PINbox.
现在将 PINbox 折叠为标准形式，如图3.22所示。图表变得简单多了，不是吗？现在您可以准确地看到新的连接是什么。设计模式实例的所有细节都已被单个 PINbox 取代。PINbox 现在充当抽象应有的角色，通过包含抽象所代表的细节并使情况更容易理解，而不是更难理解。有关使用扩展 PINbox 来显示模式内部结构的信息，请参阅第 7 章第7.1.1 节中抽象工厂的讨论。
Now collapse the PINbox to the standard form, as in Figure 3.22. The diagram just got a lot simpler, didn’t it? Now you can see precisely what the new connections are. All the details of the design pattern instance have been replaced by a single PINbox. The PINbox now acts as an abstraction should, by subsuming the details that the abstraction represents and by making the situation easier to comprehend, not harder. See the discussion of Abstract Factory in Chapter 7, Section 7.1.1, for a use of an expanded PINbox to show the internal structure of the pattern.
图 3.22. 合并的 PINbox。
Figure 3.22. Coalesced PINbox.
在这种情况下，如果您想揭示细节，您可以再次扩展 PINbox。完全移除 PINbox 并将内部与外部重新连接起来的行为称为剥离。本章开头提到的期刊文章 [ 36 ] 扩展了这些想法。
In this case, if you wanted to reveal the details, you could expand the PINbox again. The act of removing the PINbox entirely and reconnecting the internals to the externals is called peeling. The journal article [36] mentioned at the beginning of this chapter expands on these ideas.
3.3. 结论
3.3. Conclusion
PIN 旨在成为一种简单而灵活的方式，以直观的方式描述软件中的设计元素。它可以通过注释其他软件设计符号（如 UML）来使用，也可以作为独立符号来显示设计元素（如设计模式）之间的关系，而与实现需求无关。它为用户提供了几乎无限的选择，可根据特定需求在特定时间点显示哪些内容。
PIN is designed to be a simple and flexible way of visually describing design elements in software. It can be used by annotating other software design notations such as UML or as a standalone notation for showing the relationships between design elements such as design patterns independent of implementation needs. It offers the user nearly unlimited choice in degree of granularity of what can be exposed at a particular point in time, based on specific needs.
4. 与 EDP 合作
4. Working with EDPs
到目前为止，我们已经使用许多比喻（包括构建块）和与化学周期表的比较来讨论元素设计模式 (EDP)。之所以选择这些比喻和比较，是因为它们用于将基本部分组合成更大、更有用的结构至关重要的领域。这也适用于我们，并导致我们假设 EDP 可以组合成更有意义的部分。在第 3 章中，我们向您介绍了模式实例表示法 (PIN)，其中部分讨论展示了如何在图中连接 PINbox，说明了 EDP 等设计模式的多个实例如何交互。
Until now we’ve talked about Elemental Design Patterns (EDPs) using a number of metaphors, including building blocks, and comparisons to the periodic table of chemistry. Those metaphors and comparisons were selected because they are used in domains where composition of fundamental pieces into larger, more useful constructs is critical. This applies for us as well and leads to an assumption that EDPs can be joined into meaningful larger pieces. In Chapter 3, you were introduced to Pattern Instance Notation (PIN), and part of that discussion showed how PINboxes can be joined in a diagram, illustrating how multiple instances of design patterns such as EDPs can interact.
在本章中，我们将所有这些信息汇总在一起，并证明 EDP 不仅仅是小型的独立概念。EDP 构成了更大的抽象和设计模式的基础，并且可以在多种情况下有效使用，包括设计、实现和重写系统。
In this chapter, we put all of this information together and demonstrate that EDPs are not just small individual concepts. EDPs form the basis for much larger abstractions and design patterns and can be used effectively in a multitude of situations, including designing, implementing, and rewriting a system.
4.1. 形态的构成
4.1. Composition of Patterns
让我们重新回顾第 2 章第 2.2.1 节中对Decorator的讨论和解构。我们发现使用了对象递归，并且Objectifier是对象递归的一部分。解构过程需要对现有设计模式文献有深入的了解，但最终结果却很模糊，并没有让我们对Decorator有全面的了解。我们将使用 EDP 来提供更完整的画面。
Let’s revisit our discussion and deconstruction of Decorator in Chapter 2, Section 2.2.1. We identified that Object Recursion was being used and that Objectifier was a part of Object Recursion. The deconstruction process required deep knowledge of the existing design patterns literature, yet the end result was vague and did not give us a comprehensive understanding of Decorator. We’ll use EDPs to give a more complete picture.
让我们从我们刚刚描述的基本面向对象编程 EDP 之一抽象接口开始。我们声明它承诺未来的子类将为指定的方法提供实现。让我们继续使用 UML 模拟此模式的各个部分。我们知道我们有抽象接口，并且我们知道我们需要继承，因为我们讨论的是子类。我们可以如图 4.1所示显示它们。
Let’s start with one of the basic object-oriented programming EDPs we just described, Abstract Interface. We stated that it promises that a future subclass will provide an implementation for the specified method. Let’s go ahead and mock up the pieces of this pattern with UML. We know we have Abstract Interface, and we know we need Inheritance because we talk about a subclass. We can show these as in Figure 4.1.
图 4.1.抽象接口和继承EDP 作为 UML。
Figure 4.1. Abstract Interface and Inheritance EDPs as UML.
现在让我们以一种非常具体的方式连接这两个 EDP。我们知道抽象接口谈论的是Abstractor类的“未指定的子类”，这表明履行抽象器角色的类是超类。让我们通过让继承中的超类角色也指向抽象器类来表明这一点。这会将两个 UML 图合并为一个，如图4.2所示。我们应用了在第 3 章第3.2.2 节中第一次看到的内容，连接两个 EDP 实例，并创建了一个更大的设计。我们在该图中添加了另一条信息：继承的子类角色提供了来自抽象接口的操作员角色的具体定义。这是新的，在两个 EDP 中都没有出现。
Now let’s connect these two EDPs in a very specific way. We know that Abstract Interface talks about an “unspecified subclass” of the Abstractor class, suggesting that the class fulfilling the Abstractor role is a superclass. Let’s show that by having the Superclass role in Inheritance point also to the Abstractor class. This merges the two UML diagrams into one, as in Figure 4.2. We applied what we first saw in Chapter 3, Section 3.2.2, connecting the two EDP instances, and created a larger design. We added one more piece of information to this diagram: the Subclass role of Inheritance provides the concrete definition of the operator role from Abstract Interface. This is new and appears in neither EDP.
图 4.2. Fulfill Method的内部定义（UML）。
Figure 4.2. Internal definition of Fulfill Method as UML.
这个新定义满足了抽象接口的概念本质，即子类将为抽象方法提供实现，并产生了这种新设计模式的名称：Fulfill Method。这是您在本书中尚未见过的一个概念，尽管我相信如果您曾经使用面向对象语言编程，您已经见过它。我们提到了组合在第 3.2.2 节末尾，我们介绍了模式，并在图 3.12中进行了展示，但并未具体化。现在让我们来做这件事。
This new definition fulfills the contract that is the conceptual essence of Abstract Interface, that a subclass would provide an implementation for the abstract method, and gives rise to the name for this new design pattern: Fulfill Method. This is a concept you haven’t seen yet in this book, although I’m sure you’ve seen it if you’ve ever programmed in an object-oriented language. We alluded to the composition of patterns at the end of Section 3.2.2, and showed this in Figure 3.12, but it wasn’t fleshed out. Let’s do that now.
让我们重新绘制图 4.2，但不使用 UML，正如我们在 PIN 讨论中讨论的那样。在图 4.3中，实施细节被删除，该图仅显示两个 EDP 之间的关系。它清晰地抽象出每个 EDP 中包含的关系。保留一个简单的图表，更清楚地显示这两个概念之间的联系。然后，我们可以用Fulfill Method PINbox 包装这个新的简化图表，如图4.4所示。
Let’s redraw Figure 4.2 but without the UML, as we talked about in our PIN discussion. In Figure 4.3, the implementation details are removed, and the figure shows only the relationship between the two EDPs. It cleanly abstracts out the relationships that are contained within each EDP. A simple diagram remains, showing more clearly what the connection is between these two concepts. We can then wrap this new simplified diagram with the Fulfill Method PINbox, as in Figure 4.4.
图 4.3.实现方法是简单连接的 PINbox。
Figure 4.3. Fulfill Method as simple connected PINboxes.
图 4.4.履行方法作为扩展的 PINbox。
Figure 4.4. Fulfill Method as expanded PINbox.
图 4.5显示了这种表示法的真正好处，此时我们折叠展开的 PINbox 并转到更高的抽象级别。这相当于前面的每个图，但是是一个简单的 PINbox，可用于注释 UML 图，或者如我们稍后会看到的，作为更大的设计模式定义的一部分。PINbox 和设计模式的这种近乎分形的性质会在使用模式时反复出现。设计模式由较小的模式组成，其中最小的不可分割模式由 EDP 描述。反过来，从 EDP 开始的所有设计模式都可以用作更大模式的构建块。在每个粒度级别，PINbox 都可用于简洁地描述所考虑的概念。
The real payoff for this style of notation is illustrated in Figure 4.5, when we collapse the expanded PINbox and move to a higher level of abstraction. This is equivalent to each of the preceding figures but is a simple, single PINbox that can be used to annotate a UML diagram or, as we’ll see later, as a part of yet larger design pattern definitions. This almost fractal nature of PINboxes and design patterns will come up over and over again in working with patterns. Design patterns are made of smaller patterns, with the smallest, indivisible patterns described by the EDPs. In turn, all design patterns, starting with the EDPs, are usable as building blocks for larger patterns. At every level of granularity, PINboxes can be used to concisely depict the concepts under consideration.
图 4.5.履行方法作为标准 PINbox。
Figure 4.5. Fulfill Method as standard PINbox.
如果你仔细观察，你可能会注意到，我们在此过程中删除了一条信息。在最初描述图 4.2时，我说过“我们在此图中添加了一条信息：继承的子类角色提供了来自抽象接口的操作员角色的具体定义。这是新的，在两个 EDP 中都没有出现。”然而，操作的具体定义——Fulfill 方法的整个定义特征——并没有从图 4.2延续到图 4.3。
If you were following closely, you may have noticed that along the way we dropped one piece of information. When originally describing Figure 4.2, I stated that “we add one more piece of information to this diagram: the Subclass role of Inheritance provides the concrete definition of the operator role from Abstract Interface. This is new and appears in neither EDP.” The concrete definition of the operation, however—the entire defining feature of Fulfill Method—was not carried forward from Figure 4.2 to Figure 4.3.
这突出了在研究模式和使用 PIN 时必须牢记的两个要点。首先，PIN 不是模式的正式描述符号。它并非旨在定义模式的每个微小特征。它旨在让您快速轻松地处理模式实例，并在可能的情况下协助描述模式定义。如果您正在寻找设计模式的数学精确形式化，我只能再次向您指出附录作为起点。PIN 是该形式主义的近似值，旨在供人类使用。
This highlights two important points that you must keep in mind when studying patterns and using PIN. First, PIN is not a notation for the formal description of patterns. It is not designed to define every tiny feature of a pattern. It is designed to let you work with pattern instances quickly and easily and, where possible, assist in the depiction of pattern definitions. If you are looking for a mathematically precise formalization of design patterns, I can only point you once again to the appendix as a starting point. PIN is an approximation of that formalism, intended for human consumption.
其次，更重要的是，你永远不能忘记，对于任何设计模式来说，模式说明或模式规范（设计模式的散文描述）才是模式的本质和核心。数学符号可以描述模式的外观，可以列出并命名各个部分以及它们如何连接在一起，但它们永远无法告诉你关于模式的最关键的事情。它们无法告诉你为什么。它们无法告诉你何时。 或在哪里。他们最多能告诉你什么。什么是好的，但什么不是智慧。智慧，即模式旨在传授的知识，是围绕什么的一切。如果您在任何时候对模式、模式的应用方式、适用性或关键概念感到困惑，请查阅该模式的规范文档。请参阅规范。在您对模式的研究和应用中，诸如 PIN 或甚至形式演算之类的符号永远无法完全替代书写，而只能是一种助记符。
Second, and much more important, you must never forget that when it comes to any design pattern, it is the write-up or pattern specification—the prose description of a design pattern—that is the essence and heart of the pattern. Mathematical notations can describe what a pattern looks like, they can list and name the pieces and how they are hooked together, but they can never tell you the most critical thing about patterns. They cannot tell you the why. They cannot tell you the when or where. At best, they can tell you the what. What is good, but what is not wisdom. Wisdom, the knowledge that patterns are intended to impart, is everything else surrounding the what. If at any time you are confused about a pattern, about how it is being applied, about its applicability or its critical concepts, consult the canonical document for that pattern. Refer to the specification. In your studies and application of patterns, notations such as PIN or even formal calculi will never be a complete substitute for the write-up, only a mnemonic assistance.
前面的例子可能看起来并不多，但它是理解 EDP 及其如何形成更大模式的关键一步。我们取了两个 EDP，每个 EDP 定义两个实体之间的一种关系，并将它们拼接在一起形成一个稍大的模式。你刚刚看到的是良好设计的基本动作 - 选取适合当前问题的小而易懂的部分，并以有意义的方式将它们组合在一起。更重要的是，这个过程永无止境。每次我们向程序添加新的关系时，我们都会以某种方式改变设计。诀窍在于识别哪些改变是有帮助的，哪些改变会导致麻烦。
The preceding example may not seem like much, but it’s a critical step in understanding EDPs and how they form larger patterns. We took two EDPs, each of which define one relationship between two entities, and stitched them together to form a slightly larger pattern. What you just saw is the essential action of good design—taking small, understandable pieces that are appropriate to the problem at hand and putting them together in meaningful ways. What’s more, there is no end to this process. Every time we add a new relationship to a program, we alter the design in some way. The trick is recognizing which alterations are helpful and which ones lead to trouble.
在构建Fulfill Method这样一个小案例中，这一点显而易见。“当然可以，”您可能会说，“否则怎么把这两个 EDP 放在一起呢？”好吧，我们可以尝试像图 4.6那样连接它们。这是相同的两个 EDP，但是我们颠倒了哪个类是子类。现在我们无法像我们想要的那样添加该方法定义！1为了更容易看清，让我们仅使用 PINbox 重新绘制此连接，如图4.7所示。这与图 4.3完全不是同一个图。
In such a small case as constructing Fulfill Method, this may seem painfully obvious. “Well of course,” you might say, “how else would you put those two EDPs together?” Well, we could try to connect them as in Figure 4.6. These are the same two EDPs, but we reversed which class is the subclass. Now we can’t add that method definition like we wanted to!1 To make this easier to see, let’s redraw this connection with just the PINboxes, as in Figure 4.7. This is simply not the same graph as Figure 4.3.
图 4.6. 在Fulfill 方法中翻转我们的 EDP —oops。
Figure 4.6. Flipping our EDPs in Fulfill Method—oops.
图 4.7.翻转的 EDP 作为 PINbox。
Figure 4.7. Flipped EDPs as PINboxes.
只有一种特定的概念组合才能让我们达到想要的目标。这就是设计模式背后的魔力：通过反复试验，我们可以找到我们不断遇到的问题的最佳实践解决方案。我们可以将这些解决方案写成较小概念的组合，这些概念可以重复使用并塑造成新形式，同时保留原始概念连接的智慧。EDP 使我们能够从主要模块构建最佳实践解决方案，而不是通过反复试验。
Only one specific combination of concepts will get us to where we wish to be. This is the magic behind design patterns: through trial and error, we can find the best-practice solutions to problems we keep running into. We can write down these solutions as a combination of smaller concepts that can be reused and sculpted into new forms while retaining the wisdom of the original conceptual connections. EDPs allow us to build our best-practice solutions out of primary blocks instead of through trial and error.
需要说明的另一点是，图 4.2显示了满足 PINbox 中的 EDP 的绝对最小UML，而不是确切的解决方案。这不是实现；这只是一种实现。回想一下，我们提供了一种讨论设计问题（例如模式）的方式，而不会将自己限制在高度具体和严格定义的实现中。我们可以返回组成Fulfill Method的各个 EDP ，并查看它们在实现上如何有所不同，但仍然是我们希望的 EDP。
Another point to be made is that Figure 4.2 shows the absolute minimum UML that satisfies the EDPs in the PINboxes, not the exact solution. This is not the implementation; this is one implementation. Recall that we’re giving ourselves a way of talking about design issues such as patterns without limiting ourselves to a highly specific and rigidly defined implementation. We can return to the individual EDPs that comprise Fulfill Method and look at how they can differ in their implementations yet still be the EDPs we wish them to be.
4.1.1. 同位素
4.1.1. Isotopes
任何给定的设计模式，无论是小到 EDP 还是大到Decorator等成熟模式，都可以通过多种方式实现，并且仍能体现所描述的概念。这种灵活性是设计模式的标志性优势。我们将这些不同的实现称为模式的同位素。该名称旨在与术语元素一致。在化学中，同位素是特定元素的原子的变体。原子有三个组成部分：电子、质子和中子。质子和中子构成原子核，电子围绕原子核旋转。原子核中的质子数决定原子是什么元素，电子决定该原子如何与其他原子连接和相互作用。元素的同位素具有与该元素其他原子相同的质子数，并且具有相同的电子壳层结构。由于这两个特性，它在化学上的行为与该元素的其他同位素相同。它以相同的方式与其他原子连接，形成相同的键，并且在大多数情况下可以被认为与该元素的任何其他同位素原子相同。
Any given design pattern, whether one as small as an EDP or as large as a full-blown pattern such as Decorator, can be implemented in a vast number of ways and still embody the concept being described. This flexibility is a hallmark strength of design patterns. We call these differing implementations isotopes of the pattern. The name is intended to be congruent with the term elemental. In chemistry, an isotope is a variation of an atom of a particular element. Atoms have three components: electrons, protons, and neutrons. The protons and neutrons form the nucleus of the atom, and the electrons orbit the nucleus. The number of protons in the nucleus determine what element the atom is, and the electrons determine how that atom connects and interacts with other atoms. An isotope of an element has the same proton count as other atoms of that element, and it has the same electron shell structure. Because of these two traits, it behaves chemically the same as other isotopes of that element. It connects to other atoms in the same way, forms the same bonds, and for most intents and purposes can be considered the same as any other isotopic atom of that element.
然而，从内部来看，同位素原子核中的中子数量不同。从外部看，它们的化学性质和作用都相同，但内部差异可能会导致微小的副作用，例如核裂变。这往往相当具有破坏性。
Internally, however, an atom of an isotope has a different number of neutrons in the nucleus. From the outside, it looks and acts the same chemically, but the internal differences can cause small, insignificant side effects, like, oh, nuclear fission. This tends to be rather disruptive.
在软件中，我们将这种意外的核裂变称为崩溃。
In software, we call this unexpected nuclear fission a crash.
软件中设计模式的同位素定义类似。它是设计模式的一种实现，其中模式的核心概念（定义该设计模式不同于所有其他设计模式的内容）和外部接口保持不变。
An isotope of a design pattern in software is defined similarly. It is an implementation of a design pattern in which the core concepts of the pattern—what define that design pattern to be different from all other design patterns—and the external interface remain the same.
换句话说，必须履行的角色与预期的设计模式描述并无不同。2这类似于原子隐喻中的电子壳，并在 PINbox 符号中明确说明。外部实体（例如另一个模式实例或代码元素）将看到的只是定义“核心”周围的角色。
In other words, the roles that must be fulfilled do not differ from the expected design pattern description.2 This is analogous to the electron shells in the atom metaphor and is made explicit in the PINbox notation. All that an external entity, such as another pattern instance or code element, will see are the roles surrounding the “nucleus” of the definition.
原子中的质子数是区别于其他所有元素的标志，也是原子在元素周期表中的特定位置。设计模式的情况也类似，问题、解决方案和上下文都是独一无二的。设计模式的作用、解决的问题等等，都是它的鉴别器。正是它使设计模式区别于其他所有设计模式。改变问题，我们就会得到不同的设计模式。改变上下文，设计模式也必须适应并改变成其他东西。如果变化很小，而产生的新模式与原始模式非常相似，那么我们就得到了一个变体；如果变化很大，那么我们可能会得到一些全新的东西。
The proton count in an atom is what distinguishes it from all other elements and places it in a specific position in the periodic table. We have a similar situation with design patterns, where the problem, solution, and context are unique. What the design pattern does, what it solves, and so on, is its discriminator. It is what makes it unique from all other design patterns. Change the problem, and we have a different design pattern. Change the context and, again, the design pattern must adapt and change into something else. If it’s a small change, and the resulting new pattern looks a lot like the original, then we have a variant; if it’s a large change, then we may have something wholly new.
EDP 与原子和元素周期表的相似性更强，因为第 2.2.4 节中定义的设计上下文轴创建了一个明确定义的空间，我们可以在其中唯一地放置每个 EDP。这与化学元素周期表直接相似，后者创建了一个明确定义的空间，可以在其中放置元素，并且可以预测这些元素的属性。
With EDPs, the analogy to atoms and the periodic table is even stronger, because the axes of design context defined in Section 2.2.4 create a well-defined space within which we can uniquely place each EDP. This is directly comparable to chemistry’s periodic table, which creates a well-defined space in which the elements can be placed and the properties of those elements can be predicted.
EDP 在方法依赖 EDP 设计空间中的位置（由第 2.2.4 节中的对象、类型和方法相似性轴定义）是绝对的。改变这三个轴中任何一个的值，该空间中的位置都会发生变化。原始 EDP 已经突变为另一个，你可以预测新 EDP 的属性是什么。改变EDP 的任何其他内容，然而，例如它在特定实例中是如何表达的，并且 EDP 保持不变，而实现方式发生了变化——实现不同但属性相同。
The position of an EDP within the method-reliance EDP design space—as defined by the object-, type-, and method-similarity axes in Section 2.2.4—is absolute. Alter the value of any one of these three axes and the position within that space changes. The original EDP has mutated to another one, and you can predict what the properties of the new EDP will be. Change anything else about the EDP, however, such as how it is expressed in a particular instance, and the EDP stays the same while the implementation shifts—different implementation but the same properties.
同位素，无论是化学同位素还是设计模式，其不同之处在于内部结构。在原子中，中子数会发生变化。在设计模式的实例中，变化的是该实例的实现方式。这些变化可能会产生深远的影响，即使它们无法从外部立即识别出来。在任何时候，不稳定的原子都可能根据其中子数分裂，并且在任何时候，设计模式的实施不当都可能导致崩溃，即使从外部看起来没有问题。
With an isotope, either chemical or in design patterns, what differs is the internal structure. In the atom, it is the neutron count that changes. In the instance of a design pattern, it is the how that instance was implemented. These can have farranging effects, even if they are not immediately identifiable from the outside. At any moment, an unstable atom can split depending on its neutron count, and at any moment, a poorly implemented instance of a design pattern may cause a crash, even though from the outside it looks fine.
将实现与外部接口分离的想法对于面向对象编程来说并不新鲜，它非常自然且根深蒂固。同位素中引入的新概念是以相同的方式将概念的实现与接口分离。请记住，每个设计模式都有一系列角色（在参与者部分中注明），这些角色必须由实现的各个部分来满足。这些角色是设计模式与其他概念和模式交互的外部接口。
The idea of separating the implementation from the external interface is not new to object-oriented programming, it is quite natural and embedded. The new concept introduced in isotopes is separating the implementation and interface of a concept in the same manner. Remember that each design pattern has a series of roles—noted in the participants section—that must be fulfilled by portions of an implementation. These roles are the external interface through which the design pattern interacts with other concepts and patterns.
协作部分描述了这些角色在概念层面上的内部交互方式，但这些交互可以独立于实现进行描述和讨论。这就是附录中的形式化发挥作用并提供巨大灵活性的地方，但这里可以用一个代码示例来展示这个概念。我们在第 2.2.2 节中用一个方法间传递性的简单示例暗示了这一点，但现在我们可以更详细地讨论它。
The collaborations section describes how these roles interact internally at the conceptual level, but these interactions can be described and discussed independently of the implementation. This is where the formalisms from the appendix come into play and provide a tremendous amount of flexibility, but the concept can be shown here with a code example. We hinted at this back in Section 2.2.2 with a simple example of transitivity between methods, but now we can go into it in more detail.
从以下代码示例的左侧开始，其中一个对象f具有一个foo调用bar对象方法的方法b。我们称其f.foo依赖于b.bar。这两个端点之间的对象、类型和方法相似性是定义明确的，并指定了它们之间的 EDP。如果实现发生更改（例如，通过在方法调用链中注入新对象），则构造会发生变化，但端点之间的关系保持不变，如代码示例的右侧所示。原始依赖关系完好无损，三个相似性也保持完好。肯定创建了新的依赖关系，但设计模式所必需的原始依赖关系仍然存在。新实现是原始 EDP 的同位素：从外部看，它作为一个概念看起来相同且行为相同，即使内部看起来不同。
Start with the left side of the following code examples, where an object f has a method foo calling method bar of object b. We say that f.foo relies on b.bar. The object, type, and method similarities between those two endpoints is well defined and specifies the EDP between them. If the implementation is changed, for instance, by injecting a new object in a method-calling chain, the construct is altered, but the relationship between the endpoints remains unchanged, as in the right-hand side of the code example. The original reliance is intact, and the three similarities remain intact. New reliances have surely been created, but the original ones necessary for the design pattern remain. The new implementation is an isotope of the original EDP: it looks the same and acts the same when viewed as a concept from the outside, even if it looks different internally.
更重要的是，它允许开发人员讨论系统的设计，而不必关心每一个微小的实现细节。我们可以在更高的抽象层次上讨论设计。
More important, it allows developers to talk about the design of the system without having to be concerned with every tiny implementation detail. We can discuss the design at a higher level of abstraction.
举个例子，让我们重新审视Fulfill Method的构造，看看同位素如何帮助为这个小例子提供灵活性。在构造的最后，我提到图 4.2中显示的 UML是所需的最小 UML，而不是确切的要求。如果我们使用该确切的 UML定义Fulfill Method ，那么对我们实现Fulfill Method 的方式的任何更改都需要对模式进行新的定义。由于Fulfill Method在系统中出现的方式有很多，因此这将导致对如此简单的概念有大量的定义。这不是最理想的。
As an example, let’s revisit the construction of Fulfill Method and see how isotopes can help provide flexibility with even this small example. At the end of that construction, I mentioned that the UML shown in Figure 4.2 was the minimal UML required, not an exact requirement. Had we defined Fulfill Method using that exact UML, then any alteration to how we implemented Fulfill Method would require a new definition for the pattern. Because the number of ways that Fulfill Method could appear in a system are many, this would lead to a large number of definitions for such a simple concept. This is suboptimal.
在抽象接口中，充当抽象者角色的类可以有任意数量的其他字段和方法与其关联。它只需至少有一个抽象方法。图 4.8显示了多个 UML 类，其中任何一个都可以有一个与之关联的抽象接口实例。其中任何一个都可用于显示抽象接口实例的存在。类有多少个属性或字段、有多少个其他方法，或者它是模板类还是通用类，这些都无关紧要。该类只需有一个抽象方法。
The class that fulfills the Abstractor role in Abstract Interface can have any number of other fields and methods associated with it. It just has to have at least one method that is abstract. Figure 4.8 shows multiple UML classes, any one of which can have an Abstract Interface instance associated with it. Any one of these can be used to show the existence of an instance of Abstract Interface. It doesn’t matter how many attributes or fields a class has, how many other methods it may have, or whether it’s a template or generic class. The class simply has to have one method that is abstract.
图 4.8. 可以实现抽象接口EDP 的替代类。
Figure 4.8. Alternative classes that can fulfill an Abstract Interface EDP.
换句话说，图 4.4中Fulfill 方法图中显示为 PINbox的抽象接口实例可以代表其中任何一个或几乎无限数量的其他实现可能性。通过仅使用 PINbox，我们不必处理所有可能的 UML 表示或代码实现。
Stated differently, the instance of Abstract Interface shown as a PINbox in the diagram for Fulfill Method in Figure 4.4 can stand in for any of these or for a nearly infinite number of other implementation possibilities. By using just the PINboxes, we don’t have to deal with all the possible UML representations or code implementations.
同样，我们在图 4.2中看到的两类、极简版本的继承并不是满足 EDP 的唯一方法。我们感兴趣的两个类之间的继承链中可以有任意数量的类，如图4.9所示。继承树的深度无关紧要，因为子类仍然是子类。3同样，图 4.4中 PINbox 所示的继承实例代表了这些可能的实现。
Likewise, the two-class, minimalist version of Inheritance we saw in Figure 4.2 isn’t the only way to satisfy that EDP. Any number of classes can be in the inheritance chain between the two classes we’re interested in, as shown in Figure 4.9. The depth of the inheritance tree doesn’t matter because a subclass is still a subclass.3 Again, the instance of Inheritance depicted by the PINbox in Figure 4.4 is representative of any of these possible implementations.
图 4.9. 可以实现继承EDP 的替代结构。
Figure 4.9. Alternative structures that can fulfill an Inheritance EDP.
我们可以更进一步地理解这个同位素概念。回想一下，并非每种语言都有显式的类。这是第 2.2.2 节中使用术语类型的讨论的一部分。子类化如何在这些语言中工作？我们在第5 章的抽象接口规范中重新讨论了这一点，但即使是这种基于语言的重大变化也可以通过同位素来封装。记住，我们不想关心实现语言可能是什么，这就是让我们这样做的原因。通过在 EDP 和 PINbox 级别工作，我们甚至可以摆脱可能用于实现的语言的语义。语言无关性非常强大。我们在这里或多或少实现了设计概念的多态性。
We can take this isotope concept even a bit further. Recall that not every language has explicit classes. This was a component of the discussion for using the term type in Section 2.2.2. How does subclassing work in such languages? We revisit that in the specification for Abstract Interface in Chapter 5, but even that kind of significant language-based change can be encapsulated by an isotope. We do not want to have to care what the implementation language may be, remember, and this is what lets us do that. By working at the level of EDPs and PINboxes, we remove ourselves from even the semantics of the language that may be used for implementation. Being language-agnostic is extremely powerful. What we’ve accomplished here, more or less, is polymorphism of design concepts.
对于任何给定的设计模式，都有如此多的可能实现，您可能想知道我们如何知道设计模式的正确定义是什么。毕竟，我刚刚向您展示了对于任何给定的设计概念，都可能存在大量可能的实现。模式社区仍然会选择并写下一种规范形式以供共享，那么选择该规范形式的标准是什么呢？
With so many possible implementations for any given design pattern, you may be wondering how we know what the correct definition is for a design pattern. After all, I just showed you that for any given design concept, a huge number of possible implementations can exist. The patterns community still selects and writes down one canonical form for sharing, so what’s the criteria for selecting that one?
简而言之，转录的内容是最简单的实现以及参与者和协作的集合。最终形式不提供任何无关紧要的内容，而是抓住了所传达概念的本质。任何超出这个简单核心的内容都会混淆核心概念和教训。同位素几乎总是在规范描述中添加额外的项目，但仍然以这样的方式体现概念。
Simply put, what is transcribed is the simplest possible implementation and set of participants and collaborations. The final form provides nothing extraneous and gets to the essence of the concept being communicated. Anything beyond that simple core would obfuscate the central concepts and lesson. Isotopes almost always add additional items to the canonical description, yet in such a way that the concepts are still embodied.
设计模式的创建和编辑方式应抓住问题的本质、解决方案和背景，使教学、学习和使用尽可能简单。最终，最简单的描述就是最好的描述。让同位素来处理实施的修改。
Design patterns are created and edited in such a way as to get to the essence of the problem, the solution, and the context, to make teaching, learning, and using them as simple as possible. In the end, the simplest description is the best description. Let the isotopes handle the modifications of implementation.
4.2. 重新创建装饰器
4.2. Recreating Decorator
到目前为止，在本章中，您已经了解了如何将非常小的概念和设计问题以良好且具体的方式组合起来，以形成更大的概念。您还了解了这些概念如何封装实现，因此可以通过抽象简化设计讨论。我们将结合这些元素来构建一个比我们在第 2 章中开始的更令人满意和更强大的Decorator定义。
So far in this chapter, you’ve seen how very small concepts and design issues can be composed in well-formed and specific ways to form larger concepts. You’ve also seen how these concepts are an encapsulation of the implementation, so the design discussion can be simplified through abstraction. We’re going to combine these elements to build up a much more satisfying and robust definition of Decorator than we started with in Chapter 2.
我们首先重新审视图 4.10中的装饰器初始 UML 图和图 4.11中的实现方法的 UML 版本。
We start by revisiting our initial UML diagram for Decorator in Figure 4.10 and our UML version of Fulfill Method in Figure 4.11.
图 4.10.装饰器的常见示例 UML。
Figure 4.10. Decorator’s usual example UML.
图 4.11.以带注释的 UML 形式实现方法定义。
Figure 4.11. Fulfill Method definition as annotated UML.
再看看Decorator（图 4.10 ），我们可以看到Fulfill Method的结构在两个地方，Component 类在这两个地方都充当了 Abstractor 角色。再想想，这看起来很像图 2.2中的Objectifier模式，不是吗？好眼力。这正是Objectifier的本质——将单个Fulfill Method扩展到几个具体子类，如图4.12所示。同样，我们使用Fulfill Method的 PIN ，这里以堆叠形式显示（参见第 3.2.4 节），以表明底层概念是如何组合在一起的。我们刚刚定义了一个已经存在于文献中的设计模式，但从第一原理出发，通过用更小、更容易理解的部分构建它。在图 4.2到4.5中，我们将Fulfill Method包装在 PINbox 中。我们对Objectifier使用图 4.12作为内部结构，但省略了中间的图表。这为我们提供了Objectifier的 PINbox 。
Looking again at Decorator (Figure 4.10), we can see where the structure from Fulfill Method is found, in two places, with the Component class acting as the Abstractor role in both. On second thought, this is looking quite like our Objectifier pattern from Figure 2.2, isn’t it? Good eye. This is exactly what Objectifier is—the expansion of a single Fulfill Method across several concrete subclasses, as in Figure 4.12. Again, we’re using the PIN for Fulfill Method, shown here in its stacked form (see Section 3.2.4) to indicate how the underlying concepts fit together. We just defined a design pattern that already existed in the literature, but from first principles, by building it out of smaller, well-understood pieces. In Figures 4.2 through 4.5, we wrapped Fulfill Method in a PINbox. We do the same wrapping of Objectifier using Figure 4.12 as the internals but omitting the diagrams in between. This gives us a PINbox for Objectifier.
图 4.12.用 PIN 注释的对象UML。
Figure 4.12. Objectifier UML annotated with PIN.
更进一步，我们也可以定义对象递归。回想一下，我们在第 2.2.1 节中说过，对象化器是对象递归的一个组成部分，但我们没有说明剩余部分是什么。我们现在可以将其标识为受信任的重定向EDP。图 4.13将这两个模式显示为 UML。在组合形成对象递归时，FamilyHead和Objectifier将合并，就像ConcreteObjectifierB和一样Redirector。
Taking this a step further, we can define Object Recursion as well. Recall that we said in Section 2.2.1 that Objectifier is a component of Object Recursion, but we did not address what the remaining part was. We can now identify it as the Trusted Redirection EDP. Figure 4.13 shows the two patterns as UML. In combining to form Object Recursion, FamilyHead and Objectifier will merge, as will ConcreteObjectifierB and Redirector.
图 4.13.对象化器和可信重定向。
Figure 4.13. Objectifier and Trusted Redirection.
图 4.14以比原始图 2.3中介绍的形式略微简洁的形式展示了对象递归，现在用对象化器和受信任重定向的 PINbox 进行了注释。对象化器模式中的对象化器角色和受信任重定向中的家族首领角色现在由同一实体（名为的类）履行。同样，和已合并到。图 4.15中的 PIN 图更简洁地显示了这一点。HandlerConcreteObjectifierBRedirectorRecursor
Figure 4.14 shows Object Recursion in a slightly cleaner form than as introduced in the original Figure 2.3, now annotated with the PINboxes for Objectifier and Trusted Redirection. The Objectifier role from the Objectifier pattern and the Family Head role from Trusted Redirection are now being fulfilled by the same entity, the class named Handler. Likewise, ConcreteObjectifierB and Redirector have merged into Recursor. This is shown more succinctly by the PIN diagram in Figure 4.15.
图 4.14.用 PIN 注释的对象递归。
Figure 4.14. Object Recursion annotated with PIN.
图 4.15.对象递归只是 PIN。
Figure 4.15. Object Recursion as just PIN.
现在，我们可以从更小的模式角度来讨论对象递归，而且讨论方式比以前更加精确。“对象递归通过Objectifier模式使用多态性，在运行时确定相关类型系列中的哪种类型将处理特定调用。通过将受信任重定向应用于至少一种可能的实现，它还将该系列中的两个或多个对象链接在一起，以便它们可以依次处理相同的调用并发挥自己的实现作用。”这句话很准确，很直接，避免了必须从结构角度讨论模式。此外，如果有人不清楚底层概念，他或她可以研究每个子模式的完整模式规范。我们可以利用模式文献中的那些规范和定义，使我们对更高级别抽象和模式的描述更容易理解。
We can now discuss Object Recursion in terms of smaller patterns and in a much more precise way than we could before. “Object Recursion uses polymorphism, through the Objectifier pattern, to determine at runtime which type in a related family of types will handle a specific call. By applying Trusted Redirection to at least one of the possible implementations, it also chains together two or more objects from that family such that they can handle that same call in turn and bring their own implementation to bear.” This statement is precise, it is direct, and it avoids having to discuss the pattern in structural terms. Furthermore, if someone is unclear on the underlying concepts, he or she can study the full pattern specifications of each of the subpatterns. We can leverage those specifications and definitions from the patterns literature to make our descriptions of higher-level abstractions and patterns much easier to comprehend.
我们编写了Decorator模式的上半部分，只剩下最后一部分来完成我们的定义。剩下的概念是Extend Method ，它通过扩展Trusted Redirection来填充Decorator的底部，如图 4.16和4.17所示。Extend Method的原始行为和Object Recursion中的递归器合并到Decorator中，将这两个较小的模式联系在一起。4
We composed the top half of the Decorator pattern and have just one piece left to go to finish our definition. The remaining concept is Extend Method, which fills out the bottom of Decorator by extending Trusted Redirection, as shown in Figures 4.16 and 4.17. Extend Method’s original behavior and recursor from Object Recursion merge into Decorator, tying these two smaller patterns together.4
图 4.16。对象递归和扩展方法。
Figure 4.16. Object Recursion and Extend Method.
图 4.17.用 PIN 注释的装饰器。
Figure 4.17. Decorator annotated with PIN.
我们可以将其再次简化为 PINbox，如图4.18所示。此外，我们始终可以将其包装并简化为单个 PINbox，表示Decorator的一个实例，如图4.19所示。此处的角色名称直接取自四人帮 (GoF) 文本 [ 21 ] 中Decorator规范的参与者部分，并添加了两个名称：operation 和 componentObj。这些在参与者部分中隐式讨论，但我们在此将其明确化，以阐明所涉及的部分。
We can simplify this by reducing it again to just the PINboxes, as in Figure 4.18. Furthermore, we can always wrap and reduce this to a single PINbox indicating an instance of Decorator, as in Figure 4.19. The role names here are taken directly from the participants section of the Decorator specification in the Gang of Four (GoF) text [21], with two additions: operation and componentObj. These are implicitly discussed in the participants section, but we make them explicit here to clarify the pieces involved.
图 4.18.装饰器作为 PIN。
Figure 4.18. Decorator as PIN.
图 4.19.装饰器实例作为 PINbox。
Figure 4.19. Decorator instance as a PINbox.
现在，我们有了装饰器模式的简洁符号。然而，此时，我们也可以反过来。我们可以使用扩展的 PINbox，通过显示概念的底层层次结构，逐渐揭示装饰器中的更细粒度。图 4.20将装饰器显示为扩展的 PINbox，揭示了其直接的内部接线。图 4.21和4.22分别深入探讨了对象递归和对象化器，最后，图 4.23将实现方法扩展至 EDP 级别，此时我们无法进一步分解。装饰器现已完全显示出来。这些图表中的每一个都与其他图表等效。
Now we have a simple, concise notation for the Decorator pattern. At this point, however, we can go the other way as well. We can use expanded PINboxes to increasingly expose finer granularity in Decorator by showing the underlying hierarchy of concepts. Figure 4.20 shows Decorator as an expanded PINbox, revealing its direct internal wiring. Figures 4.21 and 4.22 drill into Object Recursion and Objectifier, respectively, and finally Figure 4.23 expands Fulfill Method to the EDP level, at which point we can decompose no further. Decorator is now fully revealed. Each of these diagrams is equivalent to the others.
图 4.20。扩展装饰器：一个级别。
Figure 4.20. Expanding Decorator: one level.
图 4.21。扩展装饰器：两个级别。
Figure 4.21. Expanding Decorator: two levels.
图 4.22。扩展装饰器：三个级别。
Figure 4.22. Expanding Decorator: three levels.
图 4.23。扩展装饰器：四个级别。
Figure 4.23. Expanding Decorator: four levels.
这就是Decorator，我们仅用四个 EDP 构建了它：抽象接口、继承、受信任重定向和扩展方法。每个都是一个简单的概念，但它们结合在一起，以非常具体的组合联系在一起，描述了软件系统中常见的相当高级的抽象。此外，我们证明了可以研究和掌握中间概念，以更彻底地理解Decorator。
So that’s Decorator, and we built it with just four EDPs: Abstract Interface, Inheritance, Trusted Redirection, and Extend Method. Each is a simple concept, but together, linked in a very specific combination, they describe a fairly high-level abstraction that is commonly found in software systems. What’s more, we demonstrated that interim concepts can be studied and mastered to gain a more thorough understanding of Decorator.
最重要的是，我们从未讨论过代码。我们没有提到类、方法或字段。我们只谈论概念和想法，但我们实现了一个提供指导和精确性的框架。
Most important, we never once discussed code. We didn’t bring up classes, or methods, or fields. We talked about concepts and ideas only, yet we achieved a framework that provides guidance and precision.
在 EDP 中，我们拥有构建我们理解的优秀软件的基础模块。
In EDPs, we have the building blocks with which to form great software that we understand.
GoF 文本的最后一部分 [ 21，第 358 页] 引用了 Christopher Alexander 的一句话来描述什么是“好的设计”。当我将他们自己的设计模式视为 EDP 的紧密结合和交织的例子时，我想不出比这更合适的描述。
The final section of the GoF text [21, p. 358] uses a quote from Christopher Alexander to describe what is “good design.” I can think of no more fitting description of their own design patterns when viewed as dovetailed and intertwined examples of the EDPs.
可以通过将图案以一种相当松散的方式串联起来来建造建筑物。这样建造的建筑物是图案的集合。它不是很密集。它不是很深刻。但也可以这样组合图案，使许多图案在同一物理空间中重叠：建筑物非常密集；它在小空间中捕捉到许多含义；通过这种密度，它变得深刻。[ 3，第 xli 页]
It is possible to make buildings by stringing together patterns, in a rather loose way. A building made like this, is an assembly of patterns. Is it not dense. It is not profound. But it is also possible to put patterns together in such a way that many patterns overlap in the same physical space: the building is very dense; it has many meanings captured in a small space; and through this density, it becomes profound. [3, p. xli]
设计模式文献通过密集而精确的方式组合较小的概念，充满了深刻性。EDP 让我们能够清晰而深刻地看待这种密度。
The design patterns literature is full of profundity through the composition of smaller concepts in dense and precise ways. EDPs let us view that density with clarity and insight.
还有一条评论。您还记得第 2 章开头的关于工业代码中隐藏的装饰器模式的例子吗？该模式启发了 SPQR？SPQR 使用此处描述的 EDP、组合技术和形式化方法，能够在几秒钟内识别出它的存在，而无需任何提示。与近 200 个小时相比，这可以节省很多时间。
One more comment. Do you recall the example from the beginning of Chapter 2 about the hidden Decorator pattern in industrial code that inspired SPQR? SPQR, using the EDPs, composition technique, and formalisms described here, was able to identify its existence in just a couple of seconds without being given any hints. Compared to almost 200 hours, that’s a lot of coffee breaks that could be taken instead.
4.3. 重构
4.3. Refactoring
在第 4.1.1 节讨论同位素时，我说过，一个模式同位素实例从外部看可能与该模式的另一个实例相同，但其实现存在缺陷，这会导致以后出现问题。然而，比单个实施不佳的设计模式更糟糕的是，有多个相互作用的设计模式，它们都为系统带来了善意和合理的实现选择，但结合起来却形成了问题。回想一下，第 2.1 节中描述的 SPQR 的最初驱动问题是设计模式是自然而无意地出现的。恶性设计同样容易形成，而且由于产生劣质代码的方式多于良好设计的数量，因此它们往往更容易出现。
In discussing isotopes in Section 4.1.1, I stated that a pattern isotope instance might look the same as another instance of that pattern from the outside but have a buggy implementation that causes problems later. Worse than a single poorly implemented design pattern, however, is having multiple interacting design patterns that all bring well-meaning and rational implementation choices to the system but form a problem when combined. Recall that the initial driving problem for SPQR, described in Section 2.1, was that a design pattern arose organically and unintentionally. Malignant designs can form just as easily, and because the number of ways to produce poor code is greater than the number of good designs, they tend to arise much more readily.
当谈到设计模式时，我们可以对可能出现的一些“糟糕的设计”问题进行分类。
When it comes to design patterns, we can classify some of the “bad design” issues that can appear.
首先是反模式，你可能听说过[ 14 ]。这些模式应该尽可能避免。它们代表的是常见的不良做法，而不是最佳实践。如果你的系统中存在这些模式，你应该努力将其移除，并找到合适的设计模式来解决用例。
First are the anti-patterns, which you may have heard of [14]. These are patterns that should be avoided whenever possible. They express common poor practices, not best practices. If they exist in your system, you should work to remove them and find a proper design pattern to address the use case instead.
接下来是被误用的正确设计模式。也许它们只是没有正确实施，缺少关键部分。也许设计者或实施者没有理解它们，而且它们还不够完整。这些部分设计模式通常只需进行一两个小调整即可表达原始模式。
Next up are proper design patterns that are misapplied. Perhaps they simply weren’t implemented correctly and are missing critical pieces. Perhaps they weren’t understood by the designer or implementer and aren’t quite fully formed. These partial design patterns often are just one or two small tweaks away from expressing the original pattern.
另一方面，恶性模式是指虽然发展良好，但发展却不尽如人意的模式。它们可能曾经被正确应用和实施，但随着时间的推移，代码修改已经破坏或改变了它们的形式，不再符合其最初的目的。在这种情况下，最糟糕的副作用是，文档（如果有的话）仍会经常引用它们，从而误导开发人员。
Malignant patterns, on the other hand, are patterns that grew, but grew poorly. Where they may once have been applied and implemented correctly, code modifications over time have broken or mutated them into forms that no longer fulfill their original purpose. The worst side effect in this case is that documentation, if it exists, will frequently still refer to them and thereby mislead developers.
最后，医源性模式是正确实施以解决现有问题的设计模式，因为在选择解决方案时已经了解了这一点。不幸的是，更大的设计问题带来了被忽略或未知的力量和限制。该模式解决了原始问题，但通过与其他背景力量的相互作用，无意中创造了可能比原始情况更糟糕的新情况。
Finally, iatrogenic patterns are design patterns that were implemented correctly to solve an existing problem, as it was understood when the solution was selected. Unfortunately, larger design issues brought to bear forces and constraints that were ignored or unknown. The pattern solved the original problem but, by interacting with these other context forces, inadvertently created a new situation that may be worse than the original.
所有这些问题都可以通过适当重构代码来解决，这主要需要将功能从一个位置移动到另一个位置，以更好地简化设计。系统的特性和可执行功能没有明显改变，但各部分连接在一起的方式却发生了改变。你不会直接改变代码的功能，但您可能正在增强代码的功能。通过改进系统的设计，您可以提高系统的可读性，这意味着开发人员可以更轻松地理解和修改代码。添加新功能、修复错误以及许多其他可取的操作都更简单、更快速地进行和完成。
All of these problems can be solved by properly refactoring the code, which mainly requires moving functionality from one location to another to better streamline a design. The features and executable functionality of the system are not appreciably altered, but the ways the pieces hook together are. You’re not directly altering what the code is doing, but you may be enhancing what the code could do. By improving the design of the system, you’re improving the readability of the system, which means that it is easier for a developer to understand and modify the code. Adding new features, fixing bugs, and many other desirable actions are simpler and faster to undertake and complete.
代码重构在我们这个领域已经很成熟，特别值得一提的是两个来源。Martin Fowler 的《重构》 [ 19 ] 和 Joshua Kerievsky 的《重构模式》 [ 24 ] 形成了一套将代码从一种配置转换为另一种配置的相吻合的秘诀。这两本书都值得一读。如果你曾经使用过 Eclipse、Visual Studio 或 Xcode 等集成开发环境 (IDE)，那么你可能已经通过这些工具提供的功能使用过重构。它们可以自动执行简单、有用的任务，而《重构》是大部分此类功能的起点。如果你想要深入了解如何最好地使用自动化工具，这两本书是不错的选择。
Refactoring of code is well established in our field, and two sources in particular are worth pointing out. Martin Fowler’s Refactoring [19] and Joshua Kerievsky’s Refactoring to Patterns [24] form a dovetailed set of recipes for converting code from one configuration to another. Both are well worth your time to check out. If you’ve ever used an integrated development environment (IDE) such as Eclipse, Visual Studio, or Xcode, you have probably used refactorings via the capabilities provided in those tools. They automate the process of performing simple, useful tasks, and Refactoring was the starting point for much of this functionality. If you want a solid understanding of how best to use the automated tools, you can’t go wrong with these books.
Fowler 的文本为您可以采取一系列小的、离散的重构操作奠定了基础，以改进您的代码设计。例如，他从“坏味道”开始，这些情况是即使代码可能正常运行，您也知道代码中有些地方不太对劲。他将这些直觉描述为对代码设计、代码的可理解性或表达方式的潜意识不安的可能迹象。您在编程方面的经验越多，接触到的良好设计原则越多，您就越能学会在这些情况下相信自己的直觉。
Fowler’s text lays the groundwork for a catalog of small, discrete refactoring actions that you can take to improve your code design. For instance, he starts with “bad smells,” which are situations in which you know something is not quite right with the code even though it may perform correctly. He describes these gut feelings as possible indications of a subconscious uneasiness with the code design, with how understandable or cleanly expressed it is. The more experience you have with programming and the more exposure to good design principles, the more you’ll learn to trust your instincts in these cases.
对于每一种不良气味，福勒都提供了一系列缓解措施。这些措施范围广泛且多种多样，但我在这里只关注三种。
For each bad smell, Fowler provides a series of actions to alleviate the problem. These actions are wide ranging and varied, but I focus on only three here.
例如，提取方法是一种重构操作，可用于解决涉及过长且难以理解的方法的异味。最简单的形式是，它会指导您将长方法的独立部分提取到自己的方法中并调用它。新方法的命名非常重要，这一点应该不会让您感到惊讶。
Extract Method, for example, is a refactoring action that can be used to solve a smell involving a method that’s too darned long and difficult to understand. In its simplest form, it directs you to extract a self-contained portion of the long method into its own method and call it. Great significance is given to the naming of the new method, which shouldn’t surprise you by this point.
重命名方法就体现了这一重要性，当方法的实用性或适用性可能被混淆时，可以使用该方法来澄清意图。
This significance is exemplified in Rename Method, used to clarify intent when a method’s usefulness or applicability may be obfuscated.
类似地，移动方法用于明确哪些方法属于同一类。当 Fowler 指出“某个方法正在或将要使用或被其他类的更多功能所使用，而不是定义该方法的类”时，需要进行明确说明。换句话说，当出于某种原因，某个方法位于错误的类中时，请使用移动方法。
Similarly, Move Method is used to clarify which methods belong together. Clarification is needed when, as Fowler states, “a method is, or will be, using or used by more features of another class than the class on which it is defined.” In other words, use Move Method when, for some reason, the method is in the wrong class.
现在，这些可能看起来微不足道，但每一个都有实现的微妙之处，这使得 Fowler 的贡献成为我们这个领域的经典。更多重点是，Fowler 使用这些微小、结构良好的操作作为更复杂的大型重构的构建块。Kerievsky 采用了这个概念并加以运用，使用 Fowler 的目录作为更丰富的重构的种子，展示了如何有条不紊地、精确地迁移到更常见的设计模式的实现。听起来很熟悉吧？
Now, these may seem trivial, but each one has implementation subtleties that make Fowler’s contribution necessary and rightfully a classic in our field. More to the point, Fowler uses these tiny, well-formed actions as building blocks for larger refactorings that are more complex. Kerievsky takes this concept and runs with it, using Fowler’s catalog as the seed for even richer refactorings that show how to methodically and precisely migrate toward implementations of the more common design patterns. Sound familiar?
这里有一个有趣的相似之处。我们可以通过考虑提取方法重构来继续。此重构的效果是创建一个新方法和对该方法的新调用。从中提取代码块的方法是调用站点，而新方法是被调用者。显然，这两种方法不可能相同，5因此我们在此时假设方法是不同的。此外，由于新方法是在同一个类/类型中形成的 — — 根据重构定义 — — 我们具有相同的类型和相同的对象。好吧，我们从第 2.2.4 节中的讨论中知道这是什么。具有相同对象、相同类型但不同方法的 EDP 是Conglomeration。
An interesting parallel is at work here. We can continue it by considering the Extract Method refactoring. The effect of this refactoring is to create a new method and a new call to that method. The method that the chunk of code is extracted from is the calling site, and the new method is the callee. Obviously, the two methods can’t be the same,5 so we assume at this point that the methods are dissimilar. Also, because the new method is being formed in the same class/type—according to the refactoring definition—we have the same type and also the same object. Well, we know what this is from our discussion in Section 2.2.4. The EDP with the same object, same type, but dissimilar method is Conglomeration.
接下来，让我们思考一下重命名方法，以及它对于重命名方法所涉及的 EDP 意味着什么。请记住，方法相似性（部分）由方法的名称决定。如果我们改变名称，我们可能会从方法相似的 EDP 转变为方法不相似的 EDP。例如，考虑Delegation的一个实例，如图4.24所示。如果目标方法called()重命名为caller()，它现在类似于调用站点，并且Delegation已变为Redirection，如图4.25所示。当然，重命名可以是任何名称，并且只有当新名称相似时，此 EDP 转换才成立。所以现在我们可以说Delegation +重命名方法（相似）= Redirection。反过来说，Redirection +重命名方法= Delegation。
Next, let’s ponder Rename Method and what it might mean for an EDP that the renamed method is involved in. Remember, method similarity is (partially) determined by the name of the method. If we alter the name, we could be moving from a method-similar EDP to its method-dissimilar equivalent. For example, consider an instance of Delegation, as in Figure 4.24. If the target method called() is renamed to caller(), it now is similar to the calling site, and Delegation has become Redirection, as in Figure 4.25. Of course, the rename could be to anything, and this EDP transformation only holds if the new name is a similarity match. So now we can say that Delegation + Rename Method (to similarity) = Redirection. Going the other way, Redirection + Rename Method = Delegation.
图 4.24。重命名方法重构之前的委托。
Figure 4.24. Delegation before Rename Method refactoring.
图 4.25。重命名方法重构后的委托——重定向。
Figure 4.25. Delegation after Rename Method refactoring—Redirection.
设计中的每次更改，无论多小，我们都会更改作为其组成部分的 EDP。EDP 反过来会影响由它们形成的设计模式，依此类推，直至组合层次。通过准确了解重构文献中列举的更改将如何在多个粒度级别上影响我们的设计，以及通过准确了解这些重构将如何渗透到整个系统中，我们可以更好地预测计划中的设计更改将产生哪些更大的影响。
With each change in the design, no matter how small, we alter the EDPs that are components of it. The EDPs will in turn affect the design patterns that are formed from them, and so on up the compositional hierarchy. By knowing precisely how alterations such as those enumerated in the refactoring literature will affect our design on multiple levels of granularity, and by knowing precisely how those refactorings will percolate changes throughout our system, we can better predict what the larger ramifications of a planned design change will be.
然而，不需要进行大幅度的更改就可以彻底改变设计，或引发连锁反应。我们通过应用重命名方法将委托转换为重定向，这样它就创建了以前没有的方法相似性。让我们反转这种相似性变化，但从受信任重定向开始，我们在第 4.2 节中使用它来帮助创建装饰器。你认为如果我们对受信任重定向中使用的方法执行重命名方法重构会发生什么？
Large changes aren’t required to substantially alter a design, however, or to trigger a ripple effect. We transformed Delegation to Redirection by applying Rename Method such that it created a method similarity where there wasn’t one before. Let’s reverse that similarity change but start with Trusted Redirection, which we used to help create Decorator back in Section 4.2. What do you think happens if we perform a Rename Method refactoring on the method used in Trusted Redirection?
很简单。Decorator的实例就不复存在了。
Simple. The instance of Decorator ceases to exist.
通过重命名方法，无论在实现中嵌套有多深（这因为同位素而成为可能），我们都会删除Trusted Redirection并将其替换为与方法不同的等效项Trusted Delegation。删除Trusted Redirection时，我们会删除Decorator的核心和必要部分。我们把它变成了什么？好问题。在某些方面，新的设计看起来有点像Strategy，这是 GoF 的另一种设计模式，它与Decorator在概念上有一些相似之处，但无论如何，我们的Decorator已不复存在。如果我们的文档或对系统的期望包括该实例，那么我们对系统的理解与它在实现中的实际存在方式就不匹配了。这是犯错的根源。
By renaming the method, no matter how deeply nested in the implementation—which is possible because of isotopes—we remove Trusted Redirection and replace it with its method-dissimilar equivalent, Trusted Delegation. When we remove Trusted Redirection, we remove a central and necessary piece of Decorator. What have we turned it into? Good question. In some ways, the new design looks a bit like a Strategy, another design pattern from the GoF that shares some conceptual similarities with Decorator, but in any case, our Decorator is no more. If our documentation or expectations about the system include that instance being there, we now have a mismatch between our understanding of the system and how it actually exists in implementation. That is a recipe for mistakes.
为了更有趣，让我们考虑一下移动方法，以及当它被移动时，它和调用它的任何方法之间的依赖关系会发生什么。首先，我们必须认识到移动方法意味着它将被移动到一个新的类/类型。在新类型中意味着任何调用它的方法可能必须做一些清理才能移动完成后继续使用它。让我们再次从一个使用委托的简单示例开始，如图 4.26所示。提醒一下，这是两个不同且不相关的类型的不同对象中包含的两个不同方法之间的方法调用。我们知道应用重命名方法会改变方法相似性。但是当我们移动方法时，这个 EDP 会如何变化？
For more fun, let’s consider Move Method and what happens to the reliances between it and any method that calls it when it is moved. First, we have to recognize that moving the method means that it is going to be moved to a new class/type. Being in a new type means that any method that calls it may have to do some cleanup to continue using it after the move is accomplished. Let’s start with a simple example using Delegation again, as shown in Figure 4.26. As a reminder, this is a method call between two dissimilar methods contained in two different objects of different and unrelated types. We know that applying Rename Method changes the method similarity. But how will this EDP change as we move the method around?
图 4.26。移动方法重构之前的委托。
Figure 4.26. Delegation before Move Method refactoring.
此重构具有多种可能的效果，具体取决于其实施方式。范围和影响非常广泛，现在是简要介绍另一半 EDP 的好时机。回想一下第 2.2.4 节中图 2.10中方法相似的 EDP 的二维网格。在图 4.27中，您将找到它们的方法不相似的对应物。对象类型轴被翻转以完全镜像第一个图，因为除了方法相似性之外，这些与我们之前讨论的 EDP 相同。
This refactoring has a number of possible effects, depending on how it is implemented. The range and impacts are numerous enough that now is a good time to briefly introduce the other half of the EDPs. Recall from Section 2.2.4 the two-dimensional grid of method-similar EDPs in Figure 2.10. In Figure 4.27, you’ll find their method-dissimilar counterparts. The object type axis is flipped to exactly mirror the first figure, because these are equivalent to the ones we discussed before, except for the method similarity.
图 4.27.方法相似性固定为不相似的设计空间。
Figure 4.27. The design space with method similarity fixed to dissimilar.
在移动方法时，最常见的情况是将其移动到相关类型，如图4.28所示。这样的移动会保留Delegation实例，但这并不是很有趣。考虑一下当我们将方法移动到与调用方法相同的类类型中时会发生什么，如图4.29所示。当然，现在调用called()对象obj没有多大意义。
In moving a method, the most general case is to move it to a related type, as in Figure 4.28. Such a move retains the Delegation instance, but it isn’t very interesting. Consider instead what happens when we move the method into the same class type as the calling method, as in Figure 4.29. Of course, now calling called() on the obj object doesn’t make much sense.
图 4.28。移动方法重构后的委托：无聊的情况。
Figure 4.28. Delegation after Move Method refactoring: boring case.
图 4.29。移动方法重构后的委托：变为相同类型。
Figure 4.29. Delegation after Move Method refactoring: into same type.
有两种方法可以修复方法中的不匹配问题caller()。首先，caller()可以选择将对象的类型更改obj为目标类型。换句话说，obj变为与其定义类型相同的实例字段。其次，caller()可以完全消除对象并使用调用其自己的实例self。采用哪种路径取决于数据使用需求called()。如果called()主要处理通过传递给它的数据caller()，并且该数据是封装的对象中的实例数据caller()，则可以直接访问该数据，called()而不再需要作为参数传递。在这种情况下，第二种选择是好的。否则，前一种设计是首选。
There are two possibilities for fixing the mismatch within the method caller(). First, the caller() can elect to change the type of the object obj to the destination type. In other words, obj becomes an instance field of the same type as the type it is defined in. Second, caller() can eliminate the object completely and call its own instance using self. Which path is taken depends on the data usage needs of called(). If called() works primarily on data that is passed to it via caller(), and that data is instance data in the object encapsulating caller(), then that data can be accessed by called() directly and no longer needs to be passed as parameters. In this case, the second choice is a good one. Otherwise, the former design is preferred.
但是哪些 EDP 代表了这两种情况呢？
But what EDPs represent these two cases?
第一个选择是改变对象的类型，这会导致同一类型的不同对象的不同方法之间的方法调用 EDP。通过查看图 4.27（不同对象，相似类型），我们可以看到这是一个委托聚合的示例。与重定向递归非常相似，它涉及两个相同类型的对象，协同工作。然而，与重定向递归不同，在这种情况下，这两种方法是不同的。图 4.30说明了这种情况。
The first choice, changing the type of the object, results in a method-call EDP between differing methods of differing objects of the same type. By looking at Figure 4.27, as dissimilar object, similar type, we can see that this is an example of a Delegated Conglomeration. Much like Redirected Recursion, it involves two objects of the same type, working in concert. Unlike Redirected Recursion, however, in this case the two methods are dissimilar. Figure 4.30 illustrates this situation.
图 4.30。移动方法重构后的委托：委托聚合。
Figure 4.30. Delegation after Move Method refactoring: Delegated Conglomeration.
另一种方法是简单地用 替换被调用的对象self，从而消除对字段对象的需要，如图4.31所示，并产生与之前的不同方法关系相同，但现在具有相似的类型和相似的对象。图 4.27告诉我们，这就是Conglomeration ，我们在第 2.2.4 节中讨论过它。
The alternative is to simply replace the called-upon object with self, eliminating the need for the field object, as shown in Figure 4.31, and resulting in the same dissimilar-method relationship as before, but now with a similar type, and a similar object. Figure 4.27 tells us that this is Conglomeration, which we discussed back in Section 2.2.4.
图 4.31。移动方法重构后的委托：聚集。
Figure 4.31. Delegation after Move Method refactoring: Conglomeration.
类似的二元性也出现在其他类型相关的决策中。如果要移动的方法被移动到的超类型CallingClass，那么我们要么使用受信任的委托，要么使用恢复方法，具体取决于被调用的对象是与新类型一起保留还是被删除以支持超类型。图 4.32和4.33演示了这些结果。受信任的委托非常常见，当委托要交给一组受信任的相关类型并进行适当处理时就会出现这种情况。最简单的方法是使用多态调用来调用类型为调用站点的超类的对象。超类提供了受信任的类型组，多态性意味着它得到了适当的处理。
A similar duality occurs in other type-related decisions. If the method being moved is moved to a supertype of CallingClass, then we have either Trusted Delegation or Revert Method, depending on whether the called-on object is retained with the new type or eliminated in favor of super. Figures 4.32 and 4.33 demonstrate these outcomes. Trusted Delegation is extremely common and appears when the delegation is to be handed to a trusted group of related types and handled appropriately. This is easiest to do by using a polymorphic call on an object whose type is a superclass of the calling site. The superclass provides the trusted group of types, and the polymorphism means that it is handled appropriately.
图 4.32。移动方法重构后的委托：受信任的委托。
Figure 4.32. Delegation after Move Method refactoring: Trusted Delegation.
图 4.33。移动方法重构后的委托：恢复方法。
Figure 4.33. Delegation after Move Method refactoring: Revert Method.
将方法called移至兄弟类型会导致如图 4.34CallingClass所示的情况。如您所料，这是一个委托代理的示例。在这里，通过将多态根限制为调用站点的兄弟类，可以细化可能处理类型的受信任组。
Moving the method called to a sibling type of CallingClass results in a situation such as in Figure 4.34. As you might expect, it is an example of Deputized Delegation. Here, the trusted group of possible handling types is refined by restricting the polymorphic root to a sibling class of the calling site.
图 4.34。移动方法重构后的委托：代理委托。
Figure 4.34. Delegation after Move Method refactoring: Deputized Delegation.
最后，应该注意的是，每个转换都有一个与之相反的对称转换。我们可以轻松地从我们转换为的任何 EDP返回到Delegation ，然后再返回到任何其他端点 EDP。在此过程中，我们遍历了整个图 4.27。通过对Move Method进行一次简单的重构，我们在所描述的所有六个 EDP 之间进行了路由。
Finally, it should be noted that each of the transformations has a symmetric transformation that reverses it. We could just as easily go back to Delegation from any of the EDPs we transformed into and then on to any other endpoint EDP. In doing so, we traverse the entirety of Figure 4.27. With that one simple refactoring of Move Method, we routed between all six EDPs described.
正如您所见，即使所采取的行动看似微不足道，重构也会对设计产生深远的影响。在进行重构工作时，重要的是要记住您的设计最终会走向何方，EDP 可以帮助您提前规划。
As you can see, refactoring can have far-reaching effects in a design, even when the actions taken may seem trivial. It is important to keep in mind where your design is going to end up when undertaking refactoring efforts, and EDPs can help you plan ahead.
我们可以总结到目前为止讨论过的操作和由此产生的 EDP 转换，如图4.35所示。您可以看到如何应用从委托开始的移动方法，通过粗体路径将我们引导到图表左侧的其他五个 EDP。我已经填写了其余可能的转换，包括它们的镜像转换。使用重构重命名方法向左和向右移动暗示了这里的对称性，我们将在下一节中重新讨论。
We can summarize the actions and resulting EDP transformations that we’ve discussed so far, as shown in Figure 4.35. You can see how applying Move Method, starting from Delegation, leads us to each of the other five EDPs on the left side of the diagram though the paths in bold. I’ve filled in the remaining possible transformations, including their mirror transformations. The use of the refactoring Rename Method to move left and right hints at a symmetry here, which we revisit in the next section.
图 4.35.总结迄今为止的重构效果。
Figure 4.35. Summarizing refactoring effects so far.
4.4. 总体情况
4.4. The Big Picture
到目前为止，本章已经介绍了一系列有关使用 EDP 的课程。您学习了如何将 EDP 和其他模式组合成具有可靠且可行定义的更大抽象。您学习了如何在阅读新模式时寻找您已知的模式。我们讨论了同位素，它允许设计模式及其实现在某种程度上解耦，从而允许封装设计概念并在代码中灵活表达。您甚至了解了已建立的重构方法库，了解了它们如何与 EDP 协同工作，并了解了 EDP 之间明确定义的关系如何促进重构规划。
So far, this chapter has been a series of lessons on working with EDPs. You learned how to compose EDPs and other patterns into larger abstractions that have solid and workable definitions. You learned how to look for patterns that you already know of when reading about new patterns. We discussed isotopes, which allow design patterns and their implementations to be decoupled somewhat, allowing the encapsulation of design concepts and flexibility of expression in code. You even were introduced to the established libraries of refactoring approaches, saw how they work in concert with the EDPs, and saw how the well-defined relationships among the EDPs can facilitate refactoring planning.
我们讨论了广泛的主题，但现在我们可以将它们总结成一些相当简单的图表。以下图表可以更好地了解 EDP 之间的关系。有些模式名称对您来说可能是新的。它们尚未详细讨论，但您可以在本书后面的目录章节中找到它们。
We covered a wide range of topics, but now we can summarize them into some fairly simple charts and diagrams. The following figures provide a better sense of how the EDPs relate to one another. Some pattern names will be new to you. They haven’t been discussed in detail, but you’ll find them in the catalog chapters later in this book.
图 4.36显示了哪些 EDP 被其他 EDP 使用，特别是哪些核心 EDP 涉及方法调用 EDP。如您所见，有一些清晰的安排形成了概念星座。在某些情况下，这里的用途是隐式的，而不是显式的。例如，重定向没有明确使用检索，但由于重定向在两个对象之间操作，因此必须在某个时候涉及检索，以使一个对象可供另一个对象调用其上的方法。另一方面，继承明确出现在使用子类化的六个 EDP 中。
Figure 4.36 shows which EDPs are used by other EDPs and particularly which of the core EDPs are involved in the method-call EDPs. As you can see, there are some clear arrangements that form conceptual constellations. In some cases, the uses here are implicit, not explicit. Redirection, for instance, doesn’t explicitly use Retrieve, but because Redirection operates between two objects, Retrieve must be involved at some point to make one object available to the other for calling a method on it. On the other hand, Inheritance explicitly appears in the six EDPs that use subclassing.
图 4.36. EDP 和其他选定模式之间的隐含使用关系。
Figure 4.36. Implicit used-by relationships among the EDPs and selected other patterns.
首先，请注意，图的上半部分大量使用了Retrieve 。正如您所预料的那样，使用Retrieve的 EDP涉及两个对象。这些是不同对象方法调用 EDP。它们下面是四个相似对象 EDP。
First, notice that Retrieve is used heavily in the top portion of the digram. The EDPs that use Retrieve, as you would expect, involve two objects. These are the dissimilar-object method-call EDPs. Below them are the four similar-object EDPs.
其次，继承占据了图表的中间部分。与继承相关的方法调用 EDP是那些对象类型处于子类或兄弟类类型关系的 EDP。
Second, Inheritance dominates the middle section of the diagram. Method-call EDPs involved with Inheritance are those whose object types are in a subclass or sibling class typing relationship.
还要注意，图中方法调用 EDP 从左到右是对称的。左侧是不同方法之间的 EDP；右侧是相似方法之间的 EDP。
Note also that there is a symmetry in the diagram from left to right in the method-call EDPs. On the left are the EDPs between dissimilar methods; those on the right are EDPs between similar methods.
最后，核心和方法调用 EDP 框之外的模式是由两个或多个 EDP 组成的模式。例如，我们的老朋友Fulfill Method同时使用了Abstract Interface和Inheritance 。
Finally, the patterns outside the core and method-call EDP boxes are those composed of two or more EDPs. Our old friend Fulfill Method uses both Abstract Interface and Inheritance, for instance.
我在第 2.2.4 节中向您展示了方法调用 EDP 的部分设计空间，现在在图 4.37和4.38中以图形方式进一步充实了它。每张图都显示设计空间的一半。在阅读 EDP 描述时，将其用作概念图将有助于您了解它们之间的关系。
I showed you part of the design space for method-call EDPs in Section 2.2.4, and it is now further fleshed out graphically in Figures 4.37 and 4.38. Each shows one half of the design space. Using this as a conceptual map as you read the EDP descriptions will help you see how they relate to one another.
图 4.37.完整的方法调用 EDP 设计空间：不同方法。
Figure 4.37. The full method-call EDP design space: dissimilar method.
图 4.38. 完整的方法调用 EDP 设计空间：类似方法。
Figure 4.38. The full method-call EDP design space: similar method.
这两个图对齐以匹配图 4.36的左侧和右侧。图 4.37显示了具有不同方法的所有方法调用 EDP ；它是图 4.27 （来自第 4.3 节的讨论）在图 2.9的左侧长边上的应用。通过旋转这个长方体，我们可以看到右侧长边，如图4.38所示，具有相似的方法EDP。图 4.38是图 2.10在图 2.9的右侧长边上的应用。我们在第 2.2.4 节中讨论了这些 EDP ，当我们考虑具有相似方法的 EDP 时。这两个图只是同一构造空间的左侧和右侧。它们显示了每个 EDP 在由我们开始的三个相似性轴定义的设计空间中的位置：方法、对象和对象类型。每个 EDP 都通过沿这些轴的明确变化与其周围的 EDP 相关。
These two diagrams are aligned to match with the left and right sides of Figure 4.36. Figure 4.37 shows all method-call EDPs with dissimilar methods; it is the application of Figure 4.27, from our discussion in Section 4.3, onto the left long side of Figure 2.9. By turning this cuboid, we can look at the right long side, as in Figure 4.38, with the similar-method EDPs. Figure 4.38 is the application of Figure 2.10 onto the right long side of Figure 2.9. We discussed these EDPs in Section 2.2.4, when we considered EDPs with similar methods. These two diagrams are just the left and right sides of the same constructed space. They show where each EDP sits in the design space defined by the three similarity axes we started with: method, object, and object type. Each EDP is related to the EDPs surrounding it by explicit changes along those axes.
通过描述这些变化，我们得到了图 4.39中的视图，它以图 4.35中所示为基础。该图列出了将每个方法调用 EDP 转换为另一个方法调用 EDP 所需的更改，一次一个步骤，以提供上一节介绍的原子重构。随着系统需求的变化，此图将会很有用。如果你知道你需要在这个 EDP 空间中最终达到什么位置，你就能更好地预测最终系统会是什么样子。左图对应于图 4.37，具有不同的方法，右图对应于图 4.38，具有相似的方法。如果 EDP 的方法相似性属性发生变化，则从左侧移动到右侧，反之亦然。
Describing those changes gives us the view in Figure 4.39, by building on what we showed in Figure 4.35. This lists the change required to transform each method-call EDP into another one, a single step at a time, to provide the atomic refactorings introduced in the previous section. This diagram will be useful as the requirements of a system change. You can better predict what the resulting system will look like if you know where you will need to end up in this EDP space. The figure on the left corresponds to Figure 4.37, with the dissimilar methods, and the figure on the right matches up with Figure 4.38, with the similar methods. If the method similarity property of an EDP changes, you shift from the left to the right, or vice versa.
图 4.39. 方法调用 EDP 重构关系。
Figure 4.39. Method-call EDP refactoring relations.
例如，如果您实现了递归的一个实例，但需要在多个对象之间分解任务，则标有不同对象的箭头会直接指向重定向递归。相反，如果您发现递归需要分解为不同的子任务，那么您将引入新方法，并且您知道您将不再保留方法相似性。在这种情况下，中间标有不同方法的大箭头表示您应该向左滑动到该树上的相应位置，最终到达聚合。您可以使用的每个决策点都会将您引向正确的相关概念。
For instance, if you implemented an instance of Recursion but need to break up the task among multiple objects, the arrow labeled different object points you directly to Redirected Recursion. If instead you find that the Recursion needs to be broken up into distinct subtasks, then you will be introducing new methods, and you know that you will no longer retain the method similarity. In that case, the large arrow in the middle labeled dissimilar method indicates you should slide left to the corresponding location on that tree, and you end up at Conglomeration. Each decision point that you have at your disposal will lead you to the proper, related concept.
当应用其他重构操作时，例如 Fowler [ 19 ] 和 Kerievsky [ 24 ] 所提出的操作，你会发现这些操作会沿着这些路线移动你的设计元素。通过提前了解 EDP 将如何变化和转变，你可以更好地预测你的设计最终会走向何方并管理潜在问题。
When applying other refactoring actions, such as those from Fowler [19] and Kerievsky [24], you will find those actions moving your design elements along these routes. By knowing ahead of time how the EDPs will change and shift, you can better predict where your design will end up and manage potential problems.
这四张图是理解各个 EDP 如何相互关联以及它们如何提供设计变更可能性的概念核心。在阅读以下目录时，您可能需要参考这些图以了解整体情况。虽然每个设计模式都是一个独立的概念，但它们之间存在不同的关系，形成了一个更丰富的系统，可用于将它们作为一个整体进行处理和推理。
These four diagrams are the conceptual core for understanding how the individual EDPs relate to each other and how they provide possibilities for design alterations. As you read the following catalogs, you might want to refer back to these figures to keep the larger picture in mind. While each design pattern is a self-contained concept, distinct relationships exist between them that form a richer system for working with and reasoning about them as a group.
4.5 为什么你可能想阅读附录
4.5. Why You May Want to Read the Appendix
好吧，我有点作弊了。我说过，你不必了解 EDP 背后的任何数学知识就可以理解或使用它们。不过，现在，我将根据形式主义提出一些主张，你可以你可以相信它们，也可以去阅读附录来确认它们的有效性。我希望你能保持怀疑态度并深入研究。
Okay, so I’m cheating a bit. I said that you didn’t have to understand any of the mathematics behind the EDPs to understand or use them. At this point, though, I’m going to offer a couple of claims based on the formalisms, and you can either take them on faith or you can go read the appendix to convince yourself of their validity. I hope you’re skeptical and dive in.
首先，从整体上看，EDP 的正式主体全面涵盖了面向对象编程的可能设计。请记住，本书最多只涉及现有的可能 EDP 的四分之一。如果不使用 EDP，就根本不可能编写面向对象程序。是的，我知道这是一个大胆的说法。请阅读正式的花絮以获得完整的解释。提示：我们在第 2 章中确定二元关系是我们可以形成的最小关系。给定有限数量的可以建立关系的事物，就一定存在有限数量的可能关系。考虑到我们只从对象、方法、字段和类型开始，您认为这个集合可以变得多小？如果您的答案是“大约与基本设计模式的大小相同”，那就给自己一个 cookie。这正是 EDP 的目的：一个用人类术语描述面向对象编程中所有可能的最小关系的项目，以便我们能够更有效地共享、理解和使用它们。这本书是一个开始；我希望您能加入剩余的讨论。
First off, taken as a whole, the formal body of EDPs provides complete coverage of the possible designs of object-oriented programming. Remember, this book touches on only one-quarter, at best, of the possible EDPs that exist. It is simply impossible to write an object-oriented program without using the EDPs. Yes, I know that’s a bold claim. Please read the formal tidbits for the full explanation. Hint: we established in Chapter 2 that binary relationships are the smallest relationships we can form. Given a finite number of things between which we can form relationships, there must be a finite number of possible relationships. Considering we’re starting with only objects, methods, fields, and types, how small do you think this set can get? If your answer is “about the size of the Elemental Design Patterns,” give yourself a cookie. That’s precisely what the EDPs are: a project to describe all of the possible smallest relationships in object-oriented programming in human terms so that we can share, understand, and use them more effectively. This book is a start; I hope you join in for the remainder of the discussion.
其次，ρ 演算定义的形式化是 EDP 的根本来源。大多数设计模式都是通过检查现有软件系统找到重复解决方案而创建的。这种方法让我们能够描述我们之前所做的事情，这很关键，但 EDP 的独特之处在于它们可以从第一原理定义，而不仅仅是事后定义。可以肯定的是，它们在现有系统中无处不在，这些实例是我们用来辨别 EDP 意图和编写规范的依据，就像任何其他设计模式一样。然而，由我们的正交上下文轴定义的 EDP 设计空间提供了一个框架，用于填补我们没有意识到的空白，并准确确定 EDP 之间的关系。
Second, the formalisms defined by the ρ-calculus are fundamentally where the EDPs come from. Most design patterns are created by finding repeating solutions by inspecting existing software systems. This approach lets us describe what we’ve done before, which is critical, but EDPs are unique in that they are definable from first principles, not just after the fact. To be sure, they are ubiquitously present in existing systems, and those instances are what we use to discern the intent of and write the specification for the EDP, just as with any other design pattern. The EDP design spaces defined by our orthogonal axes of context, however, provide a framework for filling in gaps we didn’t realize we had and for establishing exactly how the EDPs relate to one another.
这种差异类似于早期机械工程师共享他们通过反复试验设计的作品蓝图，而现代工程师在制造单个部件之前就能够模拟完整的设计。前者是对所尝试内容的描述，而后者的预测则基于形式推理和建模。EDP 使我们能够精确描述系统，并预测哪怕是最小的变化也会产生什么影响。此外，正如您在Decorator的重建中看到的，EDP 让我们能够根据当前需求以不同粒度级别处理大型抽象。EDP 还使我们能够描述所发现的新设计模式，并根据现有抽象编写它们的定义。可能性真的是无穷无尽的。
The difference is akin to that between early mechanical engineers sharing blueprints for creations that they designed through trial and error and modern engineers able to simulate a complete design before ever fabricating a single part. The former is a description of what was tried; the latter prediction is based on formal reasoning and modeling. EDPs enable us to describe systems precisely and predict what impact even the smallest changes will have. Furthermore, as you saw with the reconstruction of Decorator, EDPs let us work with large abstractions at varying levels of granularity based on the needs of the moment. EDPs also enable us to describe new design patterns as they are found, composing their definitions in terms of existing abstractions. There’s really no end to the possibilities.
这些特性意味着 EDP 允许我们利用软件系统做两件至关重要的事情。首先，我们可以定义和描述一个面向对象的软件，任何 面向对象软件的一部分，其设计目的是供人类使用和理解，而与实现语言无关。我们可以捕捉代码中的概念，并以适合人类推理的方式表达它们。我们表明，这些小概念可以快速构建成更大、更有趣的概念，这些概念可以使用最佳实践（例如 GoF 模式）解决实际问题。我们可以在多个粒度级别上记录我们的软件，并根据我们当时的需求选择在哪个抽象级别上进行操作。其次，更吸引人的是，由于这些概念都是形式化的，并且它们可以交互的方式同样是形式化的，因此提取、检测、组合和描述的过程可以自动化。这并非巧合，这正是 SPQR 旨在解决的问题。
These features mean that EDPs allow us to do two crucial things with software systems. First, we can define and describe a piece of object-oriented software, any piece of object-oriented software, in terms designed to be used and understood by humans, regardless of the implementation language. We can capture the concepts in the code and express them in a way that is appropriate for a human to reason about. We showed that these small concepts quickly build into larger, more interesting concepts that solve real-world problems using best practices, such as the GoF patterns. We can document our software at multiple levels of granularity and choose which level of abstraction to operate on, depending on our needs in the moment. Second, and more enticingly, because these concepts are each formalized, and the manner in which they can interact is likewise formalized, the process of extraction, detection, composition, and description can be automated. Not coincidentally, that is the problem that SPQR was designed to solve.
这仅仅是一个开始。
That’s just the beginning.
4.6. 高级主题
4.6. Advanced Topics
本节介绍了一些您可能感兴趣的高级想法。这些都不是理解 EDP 或阅读目录所必需的。如果您想浏览（但希望不要跳过）本节，请随意。
This section introduces some advanced ideas that you might find intriguing. None of these are required for understanding the EDPs or reading the catalog. If you wish to skim (but hopefully not skip) this section, please feel free to do so.
4.6.1. 重点文档和培训
4.6.1. Focused Documentation and Training
SPQR 最初旨在帮助记录现有的软件实现，让设计模式文献中描述的概念和抽象可以直接从实现中找到并记录下来。目标是公开这些设计模式实例将为开发人员提供更好的系统心理模型，并且他们将有更简单的方法来生成我们都讨厌的繁琐文档，以及更轻松地维护他们的代码库。然而，在此过程中，我们显然可能看错了方向。
SPQR was originally intended to help document existing implementations of software, letting the concepts and abstractions described by the design pattern literature be found and documented directly from the implementation. The goal was that exposing these design pattern instances would give developers a better mental model of their systems, and they would have an easier path to produce that pesky documentation that we all hate doing as well as an easier time maintaining their codebases. Somewhere along the way, however, it became obvious that maybe we were looking at the wrong thing.
设计模式是我们熟知的概念。它们有详尽的文档，很常见，是许多系统共享的基础架构。因此，如果您在软件中发现模式，那就太好了！您正在使用最佳实践来解决其他人已经解决的问题。恭喜！
Design patterns are those concepts that we understand well. They are well documented, they are common, and they are a shared infrastructure among many systems. So if you find patterns in your software, great! You are using best practices to solve problems that others have already solved. Congratulations!
但是，整个文档问题并没有解决。你仍然必须解释你是如何实现你选择的模式的，以及它们如何与系统的其余部分交互。除此之外，其他人已经解决的问题几乎肯定不是你的软件独一无二的原因。它们只是你挂载功能和巧妙的代码的骨架，这些功能和代码可以使你的系统更加吸引客户，使其在市场上脱颖而出。这些独特的部分是软件中需要花时间在文档中解决的部分。您的同事和继承您代码的人会感谢您这样做。
The entire documentation problem isn’t solved, however. You still must explain how you implemented the patterns you chose and how they interact with the rest of the system. Beyond that, however, the problems that others have solved are almost certainly not what makes your software unique. They are simply the skeleton on which you hang the features and clever bits of code that make your system more compelling to your clients, those that make it stand out in the marketplace. Those unique sections are the portions of your software that you need to spend your time addressing in documentation. Your coworkers and those who inherit your code will thank you for doing so.
换句话说，系统中可用设计模式描述的部分不一定是需要记录的关键部分。仍然有一个文档过程需要执行。但是，描述系统中哪些部分由设计模式覆盖可以告诉你哪些部分不需要花时间记录。这是代码中你不必花费过多资源培训新员工的部分，是你不需要花费数小时考虑更改的影响的部分，是你不需要确保保持精确完整以满足客户需求的部分。
In other words, the portions of your system that are describable by design patterns are not necessarily the parts that are critical to document. There’s still a documentation process to be performed. Delineating what portions of your system are covered by design patterns, however, tells you what you don’t have to spend time documenting. It’s the portion of the code that you don’t have to spend excessive resources training new hires on, the portion you don’t need to take hours to consider the impact of change to, the portion that you don’t need to make sure stays precisely intact to satisfy your customers.
高级设计模式中可描述的部分是您的基础设施，正是因为这些部分是其他人已经着手解决的部分。通过了解您的附加价值从哪里开始和结束，您可以将成本和精力范围缩小到对您的底线真正重要的部分。
The portions that are describable in high-level design patterns are those that are your infrastructure, precisely because they are portions that other people have already broken ground to solve. By understanding where your added value begins and ends, you can reduce the scope of your costs and effort to the parts that really matter to your bottom line.
当然，你还可以获得世界一流的文档，因为其他人已经写好了。你只需指出他们的工作即可。
And of course and you get world-class documentation for the remainder, because other people have already written it. You just have to point to their work.
4.6.2. 指标
4.6.2. Metrics
在任何工程学科中，定量指标都是提供反馈、确保与预测结果一致以及提供见解的必需品。软件工程也不例外，EDP 及其底层形式化使我们可以用新的方式来衡量我们所做的事情。
In any engineering discipline, quantitative metrics are a must to provide feedback, to ensure fidelity with predicted results, and to give insight. Software engineering is no different, and EDPs and their underlying formalisms make possible new ways to measure what we do.
表达清晰我们都曾遇到过可读性不如我们所愿的软件。如果幸运的话，我们可能会遇到非常容易阅读的软件。为什么会这样？为什么有些软件有意义，而另一个做同样事情的系统却很难操作？
Clarity of Expression We all have experienced software that is less readable than we would like. If we’re lucky, we may have experienced software that is extremely easy to read. Why is this? What makes some software make sense, while another system that does the same thing is a nightmare to work on?
可读性不仅仅是选择正确的名称和使用正确的空格格式的问题。如果软件能够清晰地表达其实现的概念，则被认为是可读的。当负责理解它的程序员的认知负担很低时，它就被认为是清晰的。如果软件包含正确的概念，但它们隐藏在一堆其他东西中，那么原始开发人员试图做的事情的意图和含义就很难找到。这是一个定性过程。我们一看到它就知道代码是清晰的。
Readability is not just a matter of selecting the right names and using the correct whitespace formatting. A piece of software that clearly expresses the concepts that it implements is considered readable. When the cognitive load on the programmer tasked with understanding it is low, it is considered clear. If the software contains the proper concepts, but they are hidden in a tangle of other things, then the intent and the meaning of what the original developer was trying to do is difficult to find. This is a qualitative process. We know clear code when we see it.
那么，有什么方法可以确保概念的表达清晰，并且实现尽可能易读呢？回想一下，我们讨论过同位素。我们可以通过概念表达与最小、最直接、最清晰的形式表达的匹配程度来衡量概念表达的相对清晰度。通常，这可以在特定设计模式的规范规范中找到。如果这种规范形式尚不清楚或不是最简化的形式，那么它不太可能通过社区的审查。设计模式的规范首先应该是清晰的。
So what’s one way to ensure that the expression of a concept is clear and that the implementation is as readable as possible? Well, recall that we talked about isotopes. We can measure the relative clarity of the expression of the concept by how closely it matches the smallest, most direct, clearest expression of form. Usually, this is found in the canonical specification for a particular design pattern. If this canonical form was not already clear or not at its most reduced form, it is unlikely that it could have passed muster with the community. The specification of a design pattern is intended to be, above all else, clear.
对于依赖链中的每个额外步骤或 EDP 实施中的每个不必要的部分，开发人员必须更加努力才能看到端点之间的关系。因此，我们可以通过计算对概念表达不必要的但在实施中存在的同位素组件的数量来衡量与最佳状态的距离。基本上，我们计算满足单个模式或概念实例所需的额外部分。
For each additional step in a chain of reliances or each unnecessary piece in the implementation of an EDP, the developer has to work harder to see the relationship between the end points. It follows that we can measure the distance from optimal by counting the number of isotopic components that are unnecessary to expression of the concept but are present in the implementation. Basically, we count the extra pieces required to fulfill an individual pattern or concept instance.
现在，这些额外的位对于系统中的某些其他概念来说可能是必要的和有用的，但它们的存在掩盖了特定的概念表达。我们可以用它作为代码中概念清晰度的衡量标准。我们可以测量概念的可读性，并且可以自动进行。此外，我们可以通过在 EDP 到 EDP 级别提供重构操作来指导澄清这些概念。由人类工程师来决定哪些概念最重要且需要最清楚地表达，但至少这样他们就可以采用定量方法来权衡选项并在重构计划之间进行选择。
Now, it may be (and probably is) true that those extra bits are necessary and useful for some other concept in the system, but their presence is obscuring that particular conceptual expression. We can use this as a measure of the clarity of a concept in the code. We can measure the conceptual readability, and we can do so automatically. Furthermore, we can provide guidance in clarifying these concepts by offering refactoring actions at the EDP-to-EDP level. It is up to the human engineers to decide which concepts are most important and need to be most clearly expressed, but at least this way they have a quantitative way to weigh the options and choose between refactoring plans.
抽象密度使用同位素和 EDP 为阐明意图表达提供指导，对于对代码中体现的特定概念进行细粒度精确测量非常有效，但远非全貌。代码质量的另一种衡量标准是从总体上看，获得系统的大规模、50,000 英尺视图。抽象是一种让我们将系统的各个部分视为一个单一概念的方式，而不必始终处理所有细节。如果有人告诉你一段代码是一个 for 循环，那么这比试图一步一步地描述机器级别的执行要快得多、干净得多。同样，每个设计模式或概念都是一个抽象，它用最小的带宽传达了大量的信息。此外，正如我们所展示的那样，我们可以从小的抽象中组成更大、更高级的抽象，如果能找到合适的抽象来涵盖它，我们也可以一次描述系统的相当大一部分。
Abstraction Density Using isotopes and EDPs to provide guidance for clarifying expression of intent works well for fine-grained precise measurement of specific concepts embodied in the code, but it is far from the whole picture. Another measure of code quality is to look at it in the aggregate and get a large-scale, 50,000-foot view of the system. Abstractions are a way for us to think about chunks of a system as a single concept, instead of having to deal with all details at all times. If you are told that a section of code is a for loop, that is much faster and cleaner than trying to describe the execution at the machine level, step by step. Likewise, each design pattern or concept is an abstraction that conveys a tremendous amount of information with minimal bandwidth. Further, as we showed that we can compose larger and higher-level abstractions from small ones, we can also describe quite large sections of a system at once if an appropriate abstraction can be found to encompass it.
因此，如果我们能用一个抽象而不是 100 个抽象来描述 1,000 行代码，那么我们头脑中需要管理的东西就会更少。时间。我们可以将这几千行代码作为一个整体来处理，而不必担心内部问题。可以考虑、仔细思考、理解系统的大部分内容，甚至可能进行重构。
It therefore follows that if we can describe 1,000 lines of code with one abstraction instead of 100 abstractions, we have less to manage in our heads at the same time. We can work with those thousand lines as a single whole and not have to worry about the internals. Larger sections of the system can be considered, mulled over, understood, and possibly refactored.
总体而言，在其他条件相同的情况下，能够用几个高级抽象描述的实现比必须用许多小抽象描述的实现更可取。我们称之为代码的抽象密度，以代码中存在的抽象数量为衡量标准，并根据其组成加权。换句话说，封装六个较小概念的抽象被认为比六个小抽象加上另一个小抽象更重要、更可取。在每种情况下，总共有七个抽象，但在前一种情况下，开发人员只需要执行一个抽象即可获得基本理解；在后一种情况下，必须在脑海中处理七条信息。
Overall, an implementation that can be described in a few high-level abstractions is preferable to one that must be described in many many small ones, all else being equal. We call this the abstraction density of the code, measured as the number of abstractions present in the code, weighted for their composition. In other words, an abstraction that encapsulates six smaller concepts is considered more important and preferable to the six small abstractions plus another small abstraction. In each case there are seven abstractions total, but in the former case, there’s only one that is required to be carried forward by a developer for basic understanding; in the latter situation, seven pieces of information must be mentally juggled.
上一段中的关键短语是“在其他条件相同的情况下”。抽象可以由许多信息组成，但并非所有信息都是有意义的抽象的一部分。我们可以通过测量代码中上下文无关依赖的数量来测量代码中的原始密度。请记住，每个依赖都是代码中两个实体之间的连接；因此，每个依赖都是开发人员必须跟踪的连接，但抽象减轻了这种负担。我们将这个原始数字称为代码的信息密度，以每行代码的依赖数量表示。
The key phrase in the preceding paragraph is “all else being equal.” An abstraction may be formed from many pieces of information, but not all pieces of information will be part of a meaningful abstraction. We can measure the raw density in the code by measuring the number of context-free reliances it has. Remember, each reliance is a connection between two entities in the code; therefore each reliance is a connection that the developer must keep track of, but abstractions alleviate that burden. We call this raw number the information density of the code, presented as a number of reliances per line of code.
取两者的比率使我们能够确定每单位原始信息的抽象数量，即相对抽象密度(RAD)。我们可以按每个方法、每个类、每个文件、每个目录或每个模块粒度执行此分析。它有助于通过提醒项目经理注意最有可能在未来出现问题的区域来揭示热点和潜在问题区域。这并不是说这样的代码是错误的。
Taking the ratio of the two enables us to determine the number of abstractions per unit of raw information, the relative abstraction density (RAD). We can perform this analysis on a per-method, per-class, per-file, per-directory, or per-module granularity. It helps expose hot spots and potential trouble areas by alerting a project manager to the areas that are most likely to be problematic moving forward. Which isn’t to say that such code is wrong.
RAD 越低，我们越会认为该部分代码存在问题。它可能高效、优化、运行完美且没有错误。我们只是认为当需求发生变化时，它不太容易维护。在这种情况下，也许这正是工程师需要的代码。然而，在问题出现之前了解潜在问题点在哪里，可以让您和您的团队做出相应的计划。
The lower the RAD, the more we would expect that area of the code to be problematic to work with. It may be efficient, optimized, behave perfectly, and have no bugs. We just wouldn’t consider it to be very maintainable when requirements change. Perhaps that’s exactly what an engineer needs from that code in that case. Knowing where the potential problem spots are before they bite you, however, allows you and your team to plan accordingly.
另一方面，你可能有一个模块具有非常高的 RAD，但你知道它充满了错误。你可能决定继续并消除那里的错误，因为高 RAD 让你有信心代码更易读，应该更容易使用。也许在这种情况下，培训开销足够低，你可以再雇几个人来帮忙。同样，这是一个问题将正确的信息传递到开发人员和项目负责人手中，以供人类做出决策。
On the other hand, you may have a module with a very high RAD but that you know is riddled with bugs. You may decide to go ahead and stomp the bugs there, because the high RAD gives you confidence that the code is more readable and should be easier to work with. Perhaps in that case, the training overhead is low enough that you can bring on a couple more people to help out. Again, it’s a matter of getting the right information into the hands of the developers and project leads for decision making by humans.
4.6.3. 程序分析
4.6.3. Procedural Analysis
在整个讨论过程中，一直存在一个基本假设，即所考虑的实现语言是面向对象的。毕竟，形式主义是面向对象的，UML 图是面向对象的，设计模式也是面向对象的，对吗？嗯，不完全是。设计模式只是像其他概念一样的概念。请记住，其中一些，例如创建对象和继承，可以用过程语言非常保真地表达。毕竟，那是它们最初被发现的地方。面向对象编程的出现是因为它强制执行了过程语言中使用的最佳实践。
Throughout this discussion, there has been an underlying assumption that the implementation languages under consideration are object-oriented ones. After all, the formalisms are object oriented, UML diagrams are object oriented, and design patterns are object oriented, right? Well, no, not quite. Design patterns are just concepts like any other. Remember that some of them, such as Create Object and Inheritance, can be expressed in procedural languages with great fidelity. After all, that’s where they were first found. Object-oriented programming arose because it enforced best practices that were used in procedural languages.
事实证明，只要有正确的假设，就可以像分析面向对象代码一样轻松地分析过程代码的设计模式。还记得第 2.2.2 节中关于如何在 C++ 中建模全局实体的讨论吗？在 C 中，一切都是全局的，但同样的技术也适用。在 IBM Research，我们能够成功地为包含数十万行代码的系统生成有意义的 UML 图，这些代码完全用 C 编写。EDP 是连接点。创建一组足够细粒度的形式化概念可以进行细粒度分析，以将过程习语与面向对象的假设相匹配。可视化和考虑遗留代码的全新前景开始出现。
It turns out that, given the right set of assumptions, procedural code can be analyzed for design patterns just as easily as object-oriented code can be. Remember the discussion of how to model global entities in C++, back in Section 2.2.2? In C, everything is global, but the same technique applies. At IBM Research, we were able to successfully generate meaningful UML diagrams for systems with hundreds of thousands of lines of code, written exclusively in C. EDPs were the link. Creating a fine-grained enough set of formalized concepts allows for the fine-grained analysis required to match procedural idioms to object-oriented assumptions. Entirely new vistas for visualizing and considering legacy code start to appear.
4.7. 结论
4.7. Conclusion
我相信这篇关于 EDP 的简短介绍很有趣。我希望它也能给你一些启发。软件是我们最重要的现代便利设施之一，但我们对它的掌握却极其有限。现在我们更像是炼金术士而不是工程师，在不知道元素周期表的情况下分享金块。EDP 以及催生它们的研究为更严格且非常人性化的软件工程方法奠定了基础。软件是人与机器、人与硅之间的协作，试图为其中之一进行优化不可避免地会给另一个带来问题。我建议我们让机器处理构造，我们处理概念，并使用以 EDP 为基础的设计模式作为两者之间的桥梁。
I trust this quick introduction to EDP has been interesting. I hope it has also been illuminating. Software is one of our most critical modern conveniences, one that we have an extremely tenuous grasp on. Right now we are more alchemists than engineers, sharing nuggets of gold among ourselves without knowledge of a periodic table. The EDPs, and the research that spawned them, are a way of creating a basis for a more rigorous approach to software engineering that is also very human. Software is a collaboration between man and machine, between human and silicon, and attempting to optimize it for one inevitably causes issues for the other. I propose that we let the machines handle the constructs, that we handle the concepts, and that we use design patterns with EDPs at the foundation as the bridge between the two.
现在你已经准备好阅读本书的其余部分了，我希望你在阅读时能像我在写作时一样大开眼界。如果我们作为一个社区和一个行业，可以开始从意图的角度思考软件，并专注于形成我们设计的概念，而不是我们用来向编译器解释它们的结构，我们可以专注于我们比任何机器做得更好的事情。
You’re now ready to tackle the rest of this book, and I hope you find it as much of an eye-opener in the reading as I did in the writing. If we, as a community and an industry, can start thinking of software in terms of intent and focus on the concepts that form our designs instead of on the constructs that we use to explain them to a compiler, we can concentrate on what we do better than any machine.
我们设计。我们合作。我们交流。我们解释。我们建造。
We design. We collaborate. We communicate. We explain. We build.
我们让现代世界运转起来。
We make the modern world run.
让我们将其提升至新的水平。
Let’s take it to the next level.
5. EDP 目录
5. EDP Catalog
凭借我们在前几章中建立的基础，您已准备好开始研究本书其余部分中介绍的 EDP 目录的第一部分。介绍的前 16 种模式是四种基本的面向对象编程模式，其次是十二种方法调用 EDP。到目前为止，您已经了解了其中的 11 种，还有 5 种供您根据随附的规范自行学习。
With the foundation we built in the previous chapters, you’re ready to start investigating the first portion of the EDP Catalog presented in the remainder of this book. The first sixteen patterns presented are the four fundamental object-oriented programming patterns followed by the twelve method-call EDPs. You’ve been introduced to eleven of them so far, and there are five more for you to learn on your own from the included specifications.
编写设计模式的格式是相当标准化的。最早的流行用法是在四人帮 (GoF) 的原始文本 [ 21 ] 中，我在这里也遵循这种形式。每个模式都分为几个部分，从模式名称开始，这样我们就有规范的名称来引用它。目的部分解释了模式的用途。接下来，动机部分提供了此模式解决的问题的背景，然后适用性部分解释了何时应该（或不应该）使用它。接下来提供了一个示例结构，既是 UML 图，在适当的情况下，也是扩展的 PINbox。请记住，这种结构永远不能用作死记硬背的处方，而只能作为一个例子。如果您需要更改实现，并且在不改变您所关注的关系的情况下进行更改，那么就去做吧。概念仍然存在。
The format used to write up a design pattern is fairly standardized. The first popular use was in the original Gang of Four (GoF) text [21], and I follow that form here. Each pattern is written in several sections, starting with the pattern’s name, so that we have something canonical to refer to it by. The Intent section explains the pattern’s purpose. Next, the Motivation section provides background on the problem this pattern solves, and then the Applicability section explains when it should (or should not) be used. A sample structure is provided next as both a UML diagram and, where appropriate, an expanded PINbox. This structure is not ever to be used as a rote recipe, remember, but only as an example. If you need to change the implementation and do so without altering the relationships you’re concerned with, then do it. The concepts will still be there.
参与者和协作部分紧随这些概念并加以阐述。模式规范继续介绍后果和实施部分。这里提供了示例代码，但同样，它仅供参考。您的实现语言几乎肯定会对您选择表达设计模式的方式产生很大影响。在适当的情况下，讨论这些概念如何用不同的语言表达，以强调不同的实现看起来有多么不同，但每个实现仍然可以为将底层概念作为设计模式进行一致描述提供基础。对相关模式的讨论完善了每个规范。
The Participants and Collaborations sections follow and address those concepts. The pattern specification continues with the Consequences and Implementation sections. Here, sample code is provided, but, again, it is to be taken only as a guideline. Your implementation language will almost certainly greatly impact how you choose to express a design pattern. Where appropriate, discussions of how these concepts are expressed in different languages are used to highlight just how varied different implementations may look, yet each still can provide the basis for consistent description of the underlying concepts as design patterns. A discussion of related patterns completes each specification.
EDP 分为三个主要类别：对象元素、类型关系和方法调用。
The EDPs are organized into three main categories: Object Elements, Type Relation, and Method Invocation.
对象元素是处理对象创建和定义的元素模式：创建对象和检索。创建对象描述了我们何时、如何以及为何实例化对象，它们与程序系统相比有何特殊之处，以及它们为何不仅仅是语法糖。检索概述了对象如何以及为何用作封闭对象内的数据字段。
Object Elements are those elemental patterns that deal with the creation and definition of objects: Create Object and Retrieve. Create Object describes when, how, and why we instantiate objects, what makes them special over procedural systems, and why they are not merely syntactic sugar. Retrieve outlines how and why objects are used as data fields inside an enclosing object.
类型关系包含两种简单模式：继承提供了在面向对象系统中有效重用类型信息和方法体定义的主要方法的讨论，而抽象接口解决了需要将实现推迟到可能在以后创建的类型定义的问题。
Type Relation contains two simple patterns: Inheritance provides a discussion of the primary method by which typing information and method body definitions are reused effectively in object-oriented systems, and Abstract Interface solves the problem of needing to defer implementations to type definitions that may be created at a later date.
最后一组“方法调用”包含本章的最后 12 个模式，如第 2 章所述。
The last group, Method Invocation, contains the final 12 patterns in this chapter, as described in Chapter 2.
本系列中的模式虽然小而精确，但很重要，因为它们在面向对象编程中无处不在，而且与大多数高级抽象设计模式不同，它们易于形式化。每个程序员每天都会使用这些模式，通常是无意识的。因为模式的一个目的是让人们意识到潜意识中和本能地认为有用的事物，所以这些规范旨在培养这种意识，并促进对我们领域基本概念的讨论。
The patterns in the collection, while small and precise, are important because of their ubiquity in object-oriented programming and because they are amenable to formalization, unlike most higher abstraction design patterns. Every programmer uses these patterns on a daily basis, usually without conscious effort. Because one purpose of patterns is to bring awareness to that which is subconscious and reflexively seen as useful, these specifications are intended to foster that awareness and facilitate discussion about the basic concepts of our field.
此外，EDP 还为更好地描述和讨论最佳实践设计模式奠定了基础，这些模式在当前文献中可能不那么容易理解或记忆。这些 EDP 是理解的基石，使我们能够以显著的方式重塑有关设计模式的对话。
Additionally, the EDPs act as a foundation for better describing and discussing best-practice design patterns that may not be as readily understandable or memorable from the current literature. These EDPs are the building blocks of comprehension, empowering us to reshape our conversations about design patterns in a significant way.
创建对象
Create Object
对象创建
Object Creation
意图
确保新分配的数据结构在由系统其余部分操作之前符合一组断言和先决条件，并且只能以预定义的方式进行操作。
To ensure that newly allocated data structures conform to a set of assertions and preconditions before they are operated on by the rest of the system, and that they can only be operated on in predefined ways.
也称为
实例化
Instantiation
动机
在计算机上运行的任何可执行文件中，数据必须具有存储空间，以及对数据进行操作的函数或方法。没有行为的数据无法执行操作。没有数据的行为无法执行有意义的操作。任何程序员的基本才能就是知道如何将数据和功能结合在一起，使其具有概念意义并有助于源代码的可维护性。对象实例化解决了两个问题：为数据建立默认有效状态，并为该数据建立一组相关的定义行为。
In any executable run on a computer, data must have storage space to be placed in and functions or methods to act on that data. Data without behavior is incapable of action. Behavior without data is incapable of meaningful action. A fundamental talent for any programmer is knowing how to tie together data and functionality in ways that make conceptual sense and facilitate maintainability of the source code. Object instantiation solves two issues: establishing both a default valid state for data and a set of related defined behaviors on that data.
在过程语言中，数据和行为是截然不同的，数据结构定义前者，函数处理后者。数据不仅仅是分配的内存：它是一组关于如何解释、设置和处理该内存空间中信息的假设。此外，这些假设还定义了一组对数据进行操作或使用数据有效的行为。
In procedural languages, data and behavior are distinct, with data structures defining the former and functions taking care of the latter. Data is more than just allocated memory: it is a set of assumptions about how to interpret, set, and work with the information in that memory space. In addition, those assumptions define a set of behaviors that are valid to operate on, or operate with, the data.
当需要数据时，需要占用一块内存，在许多语言中，这块内存只是被分配并交还给程序员使用。这些数据可能与某种类型相关联，也可能是无类型的原始数据。在大多数情况下，它是未初始化的，这意味着内存没有默认值。开发人员必须手动填写起始值，否则可能会有风险，即在分配之前内存位置中碰巧存在的任何随机数据仍然存在。这个过程必须由每个开发人员每次都完成。当然，封装此初始化行为的合理方法是将其放入要调用的函数中，但这样做只会将要求提高一个级别，现在开发人员必须记住调用初始化函数。不这样做意味着后面的函数对该数据进行操作时可能会输入无效或不正确的值，但此要求通常不会自动执行或严格执行。
When data is needed, it requires a block of memory to occupy, which in many languages is simply allocated and handed back to the programmer for use. This data may be associated with a type, or it may be typeless and raw. In most cases, it is uninitialized, meaning that the memory has no default value. A developer must manually fill in the beginning values or risk that whatever random data happened to be in the memory location prior to allocation is still there. This process must be done every time, by every developer. Of course, a reasonable way of encapsulating this initialization behavior is to put it into a function to be called, but doing so only moves the requirement up a level, and now the developer must remember to call the initialization function. Failing to do so means that later functions that operate on that piece of data may be fed invalid or incorrect values, but this requirement is usually not automated or strictly enforced.
清单 5.1用 C 语言演示了这种情况。假设您正在编写一个低级 GUI 库，用于为应用程序提供窗口。已经提供了一个初始化函数，但在调用它之前，如第 31 行和第 32 行所示，请求数据中的值基本上是随机的。我说“基本上”是因为 C 的一些实现确实为整数、浮点数等提供了默认值零，但这只是某些硬件上的某些实现。在其他系统上，这些值只是事先在内存中的内容。因为清单 5.1是一个完整的程序，所以您可以在您的系统上运行它并查看您的特定系统如何运行。
Listing 5.1 demonstrates this scenario in C. Assume you’re writing a low-level GUI library for offering windows to an application. An initialization function has been supplied, but before it is called, as in lines 31 and 32, the values in the requested data are essentially random. I say “essentially” because some implementations of C do provide a default value of zero for integers, floats, and such, but this is only some implementations, on some hardware. On other systems, the values are simply whatever was in memory beforehand. Because Listing 5.1 is a complete program, you can run it on your system and see how your particular system behaves.
Listing 5.1. Uninitialized data.
#include <stdio.h>
2 #include <string.h>
#include <stdlib.h>
4
struct WindowData {
6 int xPosition;
int yPosition;
8 int width;
int height;
10 char * title;
};
12
void
14 initializeWindowData( struct WindowData* wd ) {
wd->xPosition = 0;
16 wd->yPosition = 0;
wd->width = 600;
18 wd->height = 800;
// 为字符串分配足够的空间
20 //（加上 1 用于终止 NULL）
wd->title = ( char *)malloc(
22 (strlen("Default ␣ title") + 1) * sizeof ( char ));
strcpy(wd->title, "Default ␣ title");
24 };
26 int
main( int argc, char ** argv) {
28 struct WindowData wd;
30 // 这些通常打印随机数据
printf("width: ␣ %d\n", wd.width);
32 printf("title: ␣ %s\n", wd.title);
34 initialiseWindowData(&wd);
36 // 这些*总是*打印 0 和“默认标题”
printf("width: ␣ %d\n", wd.width);
38 printf("title: ␣ %s\n", wd.title);
};
#include <stdio.h>
2 #include <string.h>
#include <stdlib.h>
4
struct WindowData {
6 int xPosition;
int yPosition;
8 int width;
int height;
10 char* title;
};
12
void
14 initializeWindowData( struct WindowData* wd ) {
wd->xPosition = 0;
16 wd->yPosition = 0;
wd->width = 600;
18 wd->height = 800;
// Allocate enough for the string
20 // (Plus 1 for terminating NULL)
wd->title = (char*)malloc(
22 (strlen("Default␣title") + 1) * sizeof(char));
strcpy(wd->title, "Default␣title");
24 };
26 int
main(int argc, char** argv) {
28 struct WindowData wd;
30 // These usually print random data
printf("width:␣%d\n", wd.width);
32 printf("title:␣%s\n", wd.title);
34 initializeWindowData(&wd);
36 // These *always* print 0 and "Default title"
printf("width:␣%d\n", wd.width);
38 printf("title:␣%s\n", wd.title);
};
这导致了第二个问题，即以格式良好的方式处理格式良好的数据：不仅数据一开始就可能是错误的，而且函数对数据的运行几乎没有任何障碍。如果函数不是为处理特定数据而设计的，则可能会由于对存储在内存中的内容的假设与实际存在的内容不匹配而导致行为不正确。此外，函数对数据的错误假设可能会以无效的方式更改基础值。以前格式正确的数据现在可能对该数据类型无效。
This leads to the second issue involved with working with even well-formed data in well-formed ways: not only may the data be malformed in the beginning, but also there are few, if any, barriers to what functions may operate on it. Functions that were not designed to work on a particular piece of data may behave incorrectly due to a mismatch between assumptions of what is stored in the memory and what is actually there. In addition, a function’s incorrect assumptions about the data may alter the underlying values in an invalid way. What was previously correctly formed data may now be invalid for the data type.
通常，数据和一组相关函数会以库的形式提供，但库仅将它们作为一个松散定义的组提供；没有任何东西以系统强制执行的方式将它们主动联系在一起。作为一种改进，在某些过程语言中，可以将函数直接与它们操作的数据捆绑在一起，前提是您可以使用某种方式将函数合并到数据结构中。
Often, data and a group of related functions are shipped as a library, but a library only provides them as a loosely defined group; nothing actively ties them together in a way that is enforced by the system. As an improvement, it is possible to bundle functions directly with the data they operate on in some procedural languages, provided you have access to some way of incorporating the functions into a data structure.
在C语言中，这可以通过函数指针来实现。每个需要与结构关联的函数的函数指针都直接放在该结构中，然后与数据一起携带。1这是一种指示哪些行为与特定数据关联的好方法，但它并不能阻止开发人员直接介入并摆弄数据。
In C, this can be done with function pointers. A function pointer for each desired function to be associated with a struct is placed directly within that struct and then is carried along with the data.1 This is a good way to indicate which behavior is associated with particular data, but it doesn’t prevent a developer from simply stepping in and fiddling with the data directly.
为了保护数据，我们需要封装，即隐藏数据，同时提供一组定义明确且有限的方法来访问和操作数据。通过封装隐藏的数据被称为私有数据。封装可防止开发人员直接操作数据，并将可能的行为限制在与数据绑定的集合中。同样，只要您具有某些较低级别的访问权限（例如通过指针），就可以在过程语言中执行数据封装。此技术也称为指向实现的指针 (pimpl)、d 指针、不透明指针或柴郡猫。
To protect the data, we need encapsulation, which hides the data from view while providing a set of well-defined and limited ways to access and manipulate the data. Data hidden by encapsulation is said to be private. Encapsulation prevents a developer from directly manipulating the data and restricts the possible behaviors to the set bundled with the data. It is possible to perform data encapsulation in a procedural language, again, provided you have some lower-level access, such as through pointers. This technique is also called a pointer-to-implementation (pimpl), d-pointer, opaque pointer, or Cheshire cat.
结合这两种技术可以得到一种不错的数据保护方法，并针对该数据提供一组一致的可能行为。然而，行为捆绑和封装只在某些过程语言中可行，而且它们相当混乱、容易出错，并且需要仔细考虑和实施。更糟糕的是，他们仍然依赖这些技术作为策略应用，而没有编译器或语言的自动执行。
Combining the two techniques results in a decent way of protecting data and providing a consistent set of possible behaviors on that data. The behavior bundling and encapsulation are possible in only some procedural languages, however, and they are quite messy, error prone, and require careful consideration and implementation. Worse yet, they still rely on these techniques being applied as a matter of policy, without automated enforcement by the compiler or language.
Listing 5.2. Fixed default values.
1 struct WindowData {
int xPosition = 0;
3 int yPosition = 0;
int width = 600;
5 int height = 800;
char title[] = "默认␣标题";
7 };
1 struct WindowData {
int xPosition = 0;
3 int yPosition = 0;
int width = 600;
5 int height = 800;
char title[] = "Default␣title";
7 };
另一方面，对象是语言强制的、单一的、不可分割的单元，由数据和适用的方法组成，它们在概念上是相关的，由对象类型定义。选择作为对象类型一部分的方法已被确定为彼此之间以及与数据之间具有有意义的关联。数据可以轻松得到保护，我们可以确保在尝试操作对象之前此单元处于指定的一致且定义明确的状态，并且可以确保只能对对象执行定义明确的操作。
An object, on the other hand, is a language-enforced, single, indivisible unit composed of data and applicable methods that are conceptually related, as defined by an object type. The methods chosen to be part of the object type have been determined to have a meaningful association with each other and with the data. The data can easily be protected, we can ensure that this unit is in a specified coherent and well-defined state before we attempt to operate on the object, and we can ensure that only well-defined operations can be performed on the object.
在某些过程语言中，虽然开发人员可以很好地模拟对象以实现封装和捆绑行为，但他们无法绝对确保数据始终一致，或者对数据状态的操作受到适当限制。几乎总有某种方法可以颠覆前面描述的技术来解决这些问题。
In some procedural languages, while a developer can emulate an object reasonably well for encapsulation and bundled behaviors, they cannot absolutely ensure that the data is always coherent or that the operations on the state of that data are properly restricted. There is almost always some way to subvert the techniques described previously to address these issues.
此外，至关重要的是，从业者无法保证在分配记录时记录的内容符合他们可能选择的任何特定断言。基于我们之前的示例，我们可以为结构的成员提供默认值WindowData，如清单 5.2所示。
Also, and critically, the practitioner cannot guarantee at the time of allocation of the record that the record’s contents conform to any specific assertion they may choose to make. Building off of our earlier example, we could provide default values for the members of the WindowData struct, as in Listing 5.2.
现在我们不必调用initializeWindowData()。每当我们定义一个该类型的新变量时WindowData，它都会预先填充适当的值。
Now we don’t have to call initializeWindowData(). Whenever we define a new variable of the WindowData type, it comes prefilled with the appropriate values.
这种方法是有效的，前提是这些默认值永远不需要改变。更改这些值需要编辑和重新编译源代码。不幸的是，这通常不是我们想要的。考虑一下在大多数现代 GUI 应用程序中，当您请求新的文档窗口时，就会发生这种情况。如果存在新窗口，则新窗口会略微出现在当前窗口的右下方，并且通常标题为“无标题”，但如果同名窗口已经存在且尚未重命名，则标题会附加一个计数器，例如“无标题 1”和“无标题 2”。如果可能，库应该提供这种基本行为，因此最好在您请求新窗口时自动设置此信息WindowData。
This approach works, provided these default values never need to change. Changing the values requires editing and recompiling the source code. Unfortunately, that’s often not what we need. Consider what happens when you request a new document window in most modern GUI applications. A new window appears slightly to the right and below your current one, if one exists, and usually has the title “Untitled,” but if one with that name already exists and has not been renamed, then the title appends a counter, such as “Untitled 1” and “Untitled 2.” This sort of basic behavior should be offered by the library if possible, so it would be nice to have this information set automatically for when you request a new WindowData.
Listing 5.3. Dynamic initialization
1 void
initialiseWindowData( struct WindowData* wd ) {
3 wd->xPosition = currentWindow()->xPosition + 10;
wd->yPosition = currentWindow()->yPosition + 10;
5 wd->width = currentWindow()->width;
wd->height = currentWindow()->height;
7 char * currTitle = currentWindow()->title;
int counter = currentUntitledCounter();
9 if (counter == 0) {
// 设置为“Untitled”： 8 字符 + 1 NULL
11 wd->title =
( char *)malloc(9 * sizeof ( char ));
13 strcpy(wd->title, "Untitled");
} else {
15 // 添加足够的字符以表示 ' ' 和两位数字
// 一次限制 100 个无标题窗口
17 wd->title =
( char *)malloc(12 * sizeof ( char ));
19 strcpy(wd->title, "Untitled ␣ ");
char num[ 3 ];
21 snprintf(num, 3, "%d", counter + 1);
strcat(wd->title, num);
23 }
};
1 void
initializeWindowData( struct WindowData* wd ) {
3 wd->xPosition = currentWindow()->xPosition + 10;
wd->yPosition = currentWindow()->yPosition + 10;
5 wd->width = currentWindow()->width;
wd->height = currentWindow()->height;
7 char * currTitle = currentWindow()->title;
int counter = currentUntitledCounter();
9 if (counter == 0) {
// Set to "Untitled": 8 chars + 1 NULL
11 wd->title =
(char*)malloc(9 * sizeof(char));
13 strcpy(wd->title, "Untitled");
} else {
15 // Add enough chars for a ' ', and two digits
// Limit of 100 Untitled windows at once
17 wd->title =
(char*)malloc(12 * sizeof(char));
19 strcpy(wd->title, "Untitled␣");
char num[3];
21 snprintf(num, 3, "%d", counter + 1);
strcat(wd->title, num);
23 }
};
但我们不能这样做。刚刚描述的默认设置依赖于应用程序及其文档的现有状态，而我们在编写或编译代码时不可能知道这一点。当然，我们可以将这种动态信息收集和分配放在一个函数中，如清单 5.3 所示，但这样我们又需要开发人员记住调用这个函数。使用默认静态值并不能像我们希望的那样很好地解决我们的问题。
But we can’t do that. The default settings just described rely on the existing state of the application and its documents, and we can’t possibly know that when we’re writing or compiling the code. We can, of course, put this sort of dynamic information gathering and assignment in a function, as in Listing 5.3, but then we’re right back to requiring the developer to remember to call this function. Using default static values doesn’t solve our problem as well as we’d like.
拥有初始化函数是一种有效的解决方案，但不是可执行的。执行依赖于政策、文档和工程师纪律，但这些都未被证明是最终准确或可靠。因此，工程师又回到了最初的问题，即无法保证任何给定的断言对新分配的数据都成立。恶意、粗心或懒惰的程序员可能会分配结构，然后无法调用正确的初始化过程，从而导致潜在的灾难性后果。
Having an initializer function would be an effective solution but not an enforceable one. Enforcement relies on policy, documentation, and engineer discipline, none of which have proven to be ultimately accurate or reliable. An engineer is therefore back to the original problem of not being able to guarantee that any given assertion holds true on the newly allocated data. A malicious, careless, or lazy programmer could allocate the structure, and then fail to call the proper initialization procedure, leading to potentially catastrophic consequences.
基于对象和类的系统提供了一种替代方案。当运行时环境分配一个对象时，它会以依赖于语言和环境的格式良好的方式进行初始化。所有面向对象的环境都提供一些类似的机制作为其实现的基本部分。此机制是实现者可以创建一个函数（通常称为初始化程序或构造函数）的连接点，该函数对对象执行适当的设置。通过这种方式，可以在数据可供系统其余部分使用之前对数据施加任何特定的断言，包括那些基于仅在对象创建时可用的信息的断言。2
Object- and class-based systems provide an alternative. When an object is allocated by a runtime environment, it is initialized in a well-formed way that is dependent on the language and environment. All object-oriented environments provide some analogous mechanism as a fundamental part of their implementation. This mechanism is the attachment point at which the implementor can create a function (usually called the initializer or constructor) that performs the appropriate setup on the object. In this way any specific assertion, including those based on information only available at the time of object creation, can be imposed on the data before it is available for use by the rest of the system.2
对象用户无法绕过此机制；它由语言和运行时环境强制执行。假设的恶意、粗心或懒惰的程序员将被阻止，并避免可能的错误。由于这种类型的未初始化错误通常极难追踪和识别，因此最好避免。
There is no way for a user of the object to bypass this mechanism; it is enforced by the language and runtime environment. The hypothetical malicious, careless, or lazy programmer is thwarted, and a possible error is avoided. Because this type of missed-initialization error is generally extremely difficult to track down and identify, avoidance is preferred.
这种在程序领域中最佳实践的执行，在语言层面上提供了强大的政策执行机制，并促进了对象超越单纯的语法糖或便利性。
This enforcement of best practices in the procedural realm provides a strong policy-enforcement mechanism at the language level and promotes objects above mere syntactic sugar or convenience.
适用性
在以下情况下使用创建对象：
Use Create Object when:
• 您希望提供数据表示并强制只能在任何特定实例上执行某些操作。
• You wish to provide a data representation and enforce that only certain operations can be performed on any particular instance.
• 您希望提供数据表示的分配实例，并确保在使用之前满足一组先决条件。
• You wish to provide allocated instances of a data representation and ensure that a set of preconditions is met before use.
结构
參與者
創作網站
CreationSite
请求创建类型新对象的对象objectCreated。
The object that requests the creation of a new object of type objectCreated.
对象创建
objectCreated
要创建的对象。
The object to be created.
键入要创建的内容
TypeToCreate
的类型objectCreated。
The type of objectCreated.
合作
的实例CreationSite请求创建一个新的对象实例objectCreated，类型为TypeToCreate。执行此操作的具体机制因语言而异，但通常包括对对象类型本身发出请求。实际上，请求是向运行时环境发出的，对象类型作为参数，但大多数语言的语法都使请求看起来像是直接向对象类型发出的。敏锐的读者会注意到，这是一个自引用定义：在CreationSite创建新对象之前，必须有一个对象。启动整个链的初始对象在所用语言的运行时机制中发挥作用。例如，在 Java 中，用户告诉 JVM（Java 虚拟机）main()在哪个类中找到该方法来启动整个系统。这是一个静态方法，因此它位于类的类对象中，但它仍然是一个类对象并满足我们的要求。
An instance of CreationSite requests that a new instance of object, objectCreated, of type TypeToCreate, be created. The exact mechanism for doing so varies among languages, but it generally consists of making the request “of” the object type itself. In reality, the request is given to the runtime environment, with the object type as an argument, but the syntax of most languages creates the appearance that the request is made directly to the object type. This is, as an astute reader will notice, a self-referential definition: you must have an object to be the CreationSite before making a new object. The initial object that kicks off this whole chain comes into play in the runtime mechanism of the language being used. For instance, in Java, the user tells the JVM (Java Virtual Machine) which class to find the main() method in to start the whole system. This is a static method, so it is in the class object for the class, but, still, it’s a class object and satisfies our requirement.
结果
大多数面向对象语言不允许使用任何其他方法创建数据结构，但不需要定义开发人员提供的初始化例程。通常会提供默认初始化例程来执行最少的设置。
Most object-oriented languages do not allow for the creation of data structures using any other method, yet do not require the definition of a developer-supplied initialization routine. A default initialization routine is generally supplied that performs a minimum of setup.
有些语言虽然面向对象，但允许创建非对象数据结构。C++ 和 Objective-C 就是两个例子，它们都源自命令式 C 语言。Python、Perl 6 和其他语言也有类似的历史原因允许这种行为。
Some languages, although object oriented, allow the creation of nonobject data structures. C++ and Objective-C are two examples, both derived from the imperative C language. Python, Perl 6, and other languages have similar historical reasons for allowing such behavior.
一旦创建了对象，只有原始类的开发人员提供的一组方法才是对该对象的有效操作。请参阅继承模式以获取有关如何改变此状态的示例。
Once an object is created, only the set of methods that were supplied by the developer of the original class are valid operations on that object. See the Inheritance pattern for an example of how to alter this state of affairs.
对象在其有用存在结束时被丢弃。此释放具有类似的函数，即释放器或析构函数，它在最终释放数据的存储空间之前调用。此函数允许对对象的数据和资源施加任何后置条件。虽然为对象设置一个格式正确且明确的解构顺序也是最佳实践，但事实证明，从计算的角度来看，解构只是为了方便。有了无限的资源可供使用，对象可以继续存在，不被使用，无限期，我们永远不会意外地再次使用旧对象。只是因为我们拥有有限的资源，所以在大多数系统中，对象的销毁被认为是必要的。然而，对象的构造将它们注入到工作环境中，以便它们可用于计算，因此是一种要求，而不是一种便利。从概念上讲，某些对象类型可能依赖于对象的销毁来强制执行某些抽象概念（集合的固定元素等），但这是类设计者的问题。
Objects are disposed of at the end of their useful existence. This deallocation has an analogous function, the deallocator, or destructor, that is called before the storage space of the data is finally released. This function allows any postconditions to be imposed on the data and resources of the object. Although it is also a best practice to have a well-formed and definite deconstruction sequence for objects, it turns out that, from a computational standpoint, deconstruction is a matter of convenience only. With infinite resources at our disposal, objects could continue to exist, unused, for an indefinite amount of time, and we would never accidentally use an old one again. It is only because we have finite resources that destruction of objects is determined essential in most systems. The construction of objects, however, injects them into the working environment so they can be used in computation and is therefore a requirement, not a convenience. Conceptually, some object types may rely on the destruction of objects to enforce certain abstract notions (fixed elements of a set, etc.), but it is a matter for the class designer.
执行
在 C++ 中：
In C++:
清单 5.4是一个基本类，展示了封装、捆绑行为和默认状态。有关如何创建自己的最佳实践 C++ 类的优秀建议，请参阅 Scott Meyers 的《Effective C++》 [ 28 ] 和《More Effective C++》 [ 29 ]。
Listing 5.4 is a basic class showing encapsulation, bundled behaviors, and default state. For excellent advice on creating best practice C++ classes of your own, see Scott Meyers’s Effective C++ [28] and More Effective C++ [29].
Listing 5.4. Create Object Implementation.
class ThreadCount {
2 public:
// 构造函数
4 ThreadCount() {
numThreads = getNumberOfRunningThreads();
6 };
ThreadCount( int newData) {
8 numThreads = newData;
};
10
// 访问器
12 int getNumThreads() {
return numThreads;
14 };
ThreadCount setNumThreads( int newData) {
16 if (newData > 1) numThreads = newData; };
};
18 private:
int numThreads;
20 };
22 int
main( int argc, char ** argv) {
24 // 实例化一个 ThreadCount对象mc; 26 // 已经设置好并准备开始28 // 将会打印出正在运行的线程数 cout << mc.getPrivateData() << endl; 30 // 不会改变数据，值不好32 mc.setPrivateData(-1); 34 // 将改变数据，值正常 mc.setPrivateData(100); 36 // 打印出 100 38 cout << mc.getPrivateData() << endl; };
class ThreadCount {
2 public:
// Constructors
4 ThreadCount() {
numThreads = getNumberOfRunningThreads();
6 };
ThreadCount(int newData) {
8 numThreads = newData;
};
10
// Accessors
12 int getNumThreads() {
return numThreads;
14 };
ThreadCount setNumThreads(int newData) {
16 if (newData > 1) numThreads = newData; };
};
18 private:
int numThreads;
20 };
22 int
main(int argc, char** argv) {
24 // Instantiate an object
ThreadCount mc;
26
// Already set up and ready to go
28 // Will print out the number of running threads
cout << mc.getPrivateData() << endl;
30
// Won't change the data, value isn't good
32 mc.setPrivateData(-1);
34 // Will change the data, value okay
mc.setPrivateData(100);
36
// Prints out 100
38 cout << mc.getPrivateData() << endl;
};
相关模式
创建对象是面向对象编程的核心概念，因此随处可见。任何主要关注对象创建或分发的设计模式都基于此 EDP。示例包括下一章中的“检索新对象”和“检索共享对象”模式，以及GoF 的设计模式中的创建型模式：抽象工厂、建造者、工厂方法、原型和单例[ 21 ]。
Create Object is a core concept in object-oriented programming and as such is found everywhere. Any design pattern that concentrates mainly on the creation or distribution of objects builds off of this EDP. Examples include the Retrieve New and Retrieve Shared patterns in the next chapter and the Creational Patterns found in GoF’s Design Patterns: Abstract Factory, Builder, Factory Method, Prototype, and Singleton [21].
取回
Retrieve
对象结构
Object Structural
意图
在本地范围内使用来自另一个非本地源的对象，从而在本地范围对象和非本地源之间创建关系和连接。
To use an object from another nonlocal source in the local scope, thereby creating a relationship and a connection between the local scoping object and the nonlocal source.
动机
对象是一种成熟且易于理解的机制，用于封装通用数据和方法并执行策略，如创建对象中所示。但单个对象的实用性极其有限。事实上，如果系统中只有一个对象，而没有任何外部对象，则可以将其视为过程式程序 - 所有数据和方法都是本地的，并且完全相互公开。在任何支持使用函数对象和无方法类的面向对象系统中，都可以伪造非对象数据类型。因此，制定一种格式良好的方法来跨对象边界传输对象至关重要。在两种情况下需要这种方法，它们之间只有细微的差别。最简单的情况是外部对象具有正在访问的公开字段。更复杂的情况是外部对象具有被调用的方法，并且该方法的返回值正在内部范围内使用。
Objects are an established and well-understood mechanism for encapsulating common data and methods and enforcing policy, as shown in Create Object. Singular objects, however, are of extremely limited utility. In fact, if there were only one object in a system, and nothing external to it, it could be considered a procedural program—all data and methods are local and fully exposed to one another. Nonobject data types can be faked in any object-oriented system that supports the use of function objects and methodless classes. It is therefore critical that a well-formed methodology be put into place for transporting objects across object boundaries. There are two situations in which this methodology is necessary, and they differ only slightly. The simplest case is when an external object has an exposed field that is being accessed. The more complex case is when an external object has a method that is called, and the return value of that method is being used in the internal scope.
最简单的形式如清单 5.5所示，使用 Java 作为示例语言。
The simplest form is shown in Listing 5.5, using Java as the example language.
Listing 5.5. Retrieve with an update.
1 public class SoundSettings {
public int volume;
3 // 播放器特定的偏移量，用于调整音量
public int offset;
5 };
7 public class MusicPlayer {
public void setVolume( SoundSettings ss ) {
9 // 这是检索
此.settings.volume = ss.volume 的一个实例;
11 };
private SoundSettings settings;
13 };
1 public class SoundSettings {
public int volume;
3 // A player specific offset to adjust volume
public int offset;
5 };
7 public class MusicPlayer {
public void setVolume( SoundSettings ss ) {
9 // This is an instance of Retrieve
this.settings.volume = ss.volume;
11 };
private SoundSettings settings;
13 };
如果外部数据的使用位于临时表达式的中间，如清单 5.6adjustVolume1中的方法，则可以考虑等效方法。adjustVolume2
If the use of the external data is at the middle of a temporary expression, as in method adjustVolume1 of Listing 5.6, then an equivalent can be considered, as in method adjustVolume2.
Listing 5.6. Retrieve in a temporary variable.
1 public class MusicPlayer {
public void adjustVolume1( SoundSettings ss ) {
3 // 这也是一种检索
this .settings.volume =
5 this .settings.offset + ss.volume;
};
7 public void adjustVolume2( SoundSettings ss ) {
// 这里的检索是显式的
9 int tempVar = ss.volume;
this .settings.volume =
11 this .settings.offset + tempVar;
};
13 };
1 public class MusicPlayer {
public void adjustVolume1( SoundSettings ss ) {
3 // This is also a Retrieve
this.settings.volume =
5 this.settings.offset + ss.volume;
};
7 public void adjustVolume2( SoundSettings ss ) {
// Here the Retrieve is explicit
9 int tempVar = ss.volume;
this.settings.volume =
11 this.settings.offset + tempVar;
};
13 };
适用性
在以下情况下使用检索：
Use Retrieve when:
• 非本地对象提供对本地计算所需对象的访问，所需对象是：
• A nonlocal object provides access to an object that is required for local computation and the required object is either:
– 由方法调用的返回值提供，或者
– provided by a method call’s return value or
– 由公开的字段对象提供。
– provided by an exposed field object.
结构
请注意，这selected可以是方法，也可以是公共字段。
Note that selected could be either a method or a public field.
參與者
源访问
SourceAccess
包含选定的对象（或类）类型。
The object (or class) type that contains selected.
水槽支架
SinkHolder
包含要赋予新值的项目target 的对象（或类）类型。
The object (or class) type that includes the item, target, to be given a new value.
检索
Retrieved
要更新的值的类型和返回的值。
The type of the value to be updated and the value that is returned.
下沉
sink
被赋予新值的字段。
The field that is given a new value.
来源
source
产生新值的方法或字段。
The method or field that produces the new value.
合作
这种简单的关系由两个对象和两个方法组成。区别因素是将返回值传输到本地对象空间，并使用检索到的对象更新本地字段。本地字段可能是已定义的字段，也可能是表达式中间的临时值。
This simple relationship consists of two objects and two methods. The distinguishing factors are the transferral of a return value into the local object space and an update to a local field using that retrieved object. The local field may be a defined field, or it could be a temporary value in the middle of an expression.
结果
像这样绑定两个对象和/或类型是日常现象，但不应不假思索地这样做。任何时候以这种方式绑定两个对象，您都会引入依赖关系：目标对象现在依赖于源对象。确保这是您需要做的。
Tying two objects and/or types like this is an everyday occurrence, but it should not be done without thought. Any time you bind two objects in this manner, you are introducing a dependency: the target object now relies on the source object. Make sure it’s what you need to do.
在具有动态类型的语言（例如 Python、Ruby 或 JavaScript）中，SourceAccess、SinkHolder 和 Retrieved 的类型角色可能并非在所有情况下都明确显示。此外，开发人员可能会或可能不会提供强类型。
In languages with dynamic typing, such as Python, Ruby, or JavaScript, the type roles of SourceAccess, SinkHolder, and Retrieved may not be explicitly shown in all cases. Also, a developer may or may not provide strong typing.
执行
在 C++ 中：
In C++:
1 class SourceAccess {
public:
3 检索到的源();
};
5
class SinkHolder {
7 检索到的接收器;
SourceAccess srcobj;
9 public:
void operation() { sink = srcobj.source(); }
11 };
1 class SourceAccess {
public:
3 Retrieved source();
};
5
class SinkHolder {
7 Retrieved sink;
SourceAccess srcobj;
9 public:
void operation() { sink = srcobj.source(); }
11 };
在 Java 中：
In Java:
1 public class SourceAccess {
public Retrieved source();
3 };
5 public class SinkHolder {
private Retrieved sink;
7 private SourceAccess srcobj;
public void operation() {
9 sink = srcobj.source();
};
11 };
1 public class SourceAccess {
public Retrieved source();
3 };
5 public class SinkHolder {
private Retrieved sink;
7 private SourceAccess srcobj;
public void operation() {
9 sink = srcobj.source();
};
11 };
在 Python 中（注意动态类型）：
In Python (note dynamic typing):
1 class SourceAccess(object):
def source(self):
3 pass // 在这里返回值
5 class SinkHolder(object):
def void operation(self):
7 self.sink = self.srcobj.source()
1 class SourceAccess(object):
def source(self):
3 pass // Return value here
5 class SinkHolder(object):
def void operation(self):
7 self.sink = self.srcobj.source()
相关模式
检索是一种基本的 EDP，可以与涉及两个或多个对象的其他模式一起使用，这些对象在运行时被组合在一起。请参阅第 6 章中的检索新对象和检索共享对象模式，了解此 EDP 的变体，其中涉及对象所有权和实例化。
Retrieve is a fundamental EDP and is found with any other pattern that involves two or more objects, where those objects are brought together at runtime. See the Retrieve New and Retrieve Shared patterns in Chapter 6 for variations on this EDP where object ownership and instantiation are involved.
遗产
Inheritance
类型关系
Type Relation
意图
重用另一个类的接口、实现和行为，并对其中每个进行添加或修改。
To reuse another class’s interface, implementation, and behavior with additions to or alterations of each.
也称为
IsA，类型重用
IsA, Type Reuse
动机
通常，现有类为创建新类类型提供了良好的开端。接口可能几乎就是您正在寻找的，现有方法可能几乎提供了新类所需的一切，或者至少现有类在概念上可能接近您想要实现的目标。
Often, an existing class provides an excellent start for producing a new class type. The interface may be almost exactly what you are looking for, the existing methods may provide almost what you need for your new class, or, at the very least, the existing class may be conceptually close to what you wish to accomplish.
在这种情况下，重用现有类而不是从头开始重写所有内容是有用且高效的。这样做的一种方法是将代码复制并粘贴到新类中。这种技术经常使用，但它有很多缺点。如果在原始代码中发现错误，您现在有两个地方可以跟踪和维护修复。如果对代码的一个副本进行了增强，则必须明确告知在另一个代码位置上工作的任何人它存在。复制和粘贴一开始似乎是一种快速简便的方法，但它不仅将代码副本联系在一起，还将开发团队联系在一起。此外，在许多情况下，您根本没有要复制的源代码，例如在使用供应商的开发 API 时。
In such cases, it is useful and efficient to reuse the existing class instead of rewriting everything from scratch. One way of doing so is by copying and pasting the code into a new class. This technique is done quite often, but it has many drawbacks. If a bug is found in the original code, you now have two places to track and maintain the fix. If an enhancement is made to one copy of the code, anyone working on the other code location must be explicitly told that it exists. Copy and paste seems like a quick and easy approach in the beginning, but it doesn’t just tie the copies of code together, it ties the development teams together. In addition, in many cases, you won’t have the source code to copy at all, such as when using a vendor’s development API.
实现这种重用的更好方法是通过继承模式。每种面向对象语言都支持这种代码重用方法，它通常是此类语言的核心原语。
A better way to accomplish this reuse is through the Inheritance pattern. Every object-oriented language supports this approach of code reuse, and it is usually a core primitive of such a language.
从最基本的层面上讲，此模式提供了超类或基类与子类或派生类之间的关系。超类（我们称之为Superclass）是系统中现有的类，它至少为方法和/或数据结构提供了概念接口。第二个类可以定义为派生自；Superclass我们称之为Subclass。Subclass该类继承了接口和所有方法和字段的实现Superclass，这为程序员开始工作提供了一个起点Subclass。
At its most basic level, this pattern offers a relationship between a superclass or base class, and a subclass or derived class. The superclass (let’s call it Superclass) is an existing class in the system, one that provides at a minimum an interface of concepts for methods and/or data structures. A second class can be defined as being derived from Superclass; let’s call it Subclass. The Subclass class inherits the interface and implementations of all methods and fields of Superclass, and this provides a starting point for a programmer to begin work on Subclass.
假设您正在编写一个库来在图形界面中显示形状，例如游戏引擎的核心。无论形状是什么类型，每个形状都会有一些基本信息。它在屏幕上会有一个位置，会有一种颜色，绘制其边框的线条会有一定的厚度，等等。您可以将这些信息添加到每个表示形状的类中，并实现处理数据的方法。或者，您可以使用继承，如清单 5.7所示。该类Shape为系统中的任何形状设置基本信息，并提供一组用于处理该数据的基本方法实现。
Assume that you are writing a library to display shapes in a graphical interface, such as might be at the core of a game engine. Every shape is going to have some basic information, regardless of what kind of shape it is. It will have a position on the screen, it will have a color, the line that draws its border will have a thickness, and so on. You could add this information, and implement the methods to work with the data, to each and every class that represents a shape. Or, you could use Inheritance as in Listing 5.7. The Shape class sets up the basic information for any shape in the system and provides a basic set of method implementations for working with that data.
你可能注意到Square我们在类中重新声明了实例方法setWidth:andHeight。这是重写和是与继承一起出现的常见行为。通过提供方法的新定义，覆盖可让您自定义基类中的某些方法。您可以使用它来更改现有方法的行为，要么完全替换它，要么以新的方式扩充它。清单 5.8演示了使用覆盖Square。我们希望让和的客户端Square仍然Rectangle使用setWidth:andHeight接口，因为正方形只是一种特殊的矩形，但现在我们必须确保边长相同。我们添加对数据的检查，然后从超类中调用预先存在的方法。没有必要完全替换实现，因为我们只是将其包装在有效性检查中。3
You may notice that in the Square class we re-declared the instance method setWidth:andHeight. This is an example of overriding and is a behavior that commonly goes along with Inheritance. Overriding lets you customize certain methods from a base class by providing a new definition of the method. You can use it to change the behavior of an existing method, either by completely replacing it or by augmenting it in new ways. Listing 5.8 demonstrates overriding with Square. We’d like to let clients of both Square and Rectangle still use the setWidth:andHeight interface, since a square is just a special kind of rectangle, but now we have to make sure the sides are the same size. We add a check on the data, and then call the preexisting method from our superclass. There’s no need to completely replace the implementation because we’re just wrapping it in a validity check.3
Listing 5.7. Basic inheritance example in Objective-C.
1 @interface Shape
{
3 int xPos;
int yPos;
5 int lineWidth;
Color* fillColor;
7 }
- ( void ) setPosWithX: ( int ) x andY: ( int ) y;
9 - ( void ) setColor: (Color*) c;
- ( void ) setLineWidth: ( int ) lw;
11 @end
13 @interface Circle : Shape
{
15 int radius;
}
17 - ( void ) setRadius: ( int ) r;
@end
19
@interface Rectangle : Shape
21 {
int width;
23 int height;
}
25 - ( void ) setWidth: ( int ) w andHeight: ( int ) h;
@end
27
@interface Square : Rectangle
29 {
}
31 - ( void ) setWidth: ( int )w andHeight: ( int ) h;
- ( void ) setSize: ( int ) s;
33 @end
1 @interface Shape
{
3 int xPos;
int yPos;
5 int lineWidth;
Color* fillColor;
7 }
- (void) setPosWithX: (int) x andY: (int) y;
9 - (void) setColor: (Color*) c;
- (void) setLineWidth: (int) lw;
11 @end
13 @interface Circle : Shape
{
15 int radius;
}
17 - (void) setRadius: (int) r;
@end
19
@interface Rectangle : Shape
21 {
int width;
23 int height;
}
25 - (void) setWidth: (int) w andHeight: (int) h;
@end
27
@interface Square : Rectangle
29 {
}
31 - (void) setWidth: (int)w andHeight: (int) h;
- (void) setSize: (int) s;
33 @end
Listing 5.8. Overriding an implementation.
1 @implementation Square
- ( void ) setWidth: ( int ) w andHeight: ( int ) h
3 {
if (w == h) {
5 [ super setWidth: w andHeight: h];
} else {
7 printf("ERR: ␣ Width ␣ != ␣ height\n");
}
9 }
- ( void ) setSize: ( int ) s
11 {
[ self setWidth: s andHeight: s];
13 }
@end
1 @implementation Square
- (void) setWidth: (int) w andHeight: (int) h
3 {
if (w == h) {
5 [super setWidth: w andHeight: h];
} else {
7 printf("ERR:␣Width␣!=␣height\n");
}
9 }
- (void) setSize: (int) s
11 {
[self setWidth: s andHeight: s];
13 }
@end
有时您需要保持依赖原始行为的代码的完整性，但又想为新代码提供一个修复版本。您可以通过创建一个新类来实现这一点，该新类继承自原始类，但会覆盖损坏的方法并提供修复版本。现有代码可以继续使用基类，并解决可能正在解决的错误，但新代码可以利用修复后的版本。在对旧代码进行检查和测试后，也可以将其迁移到新版本。
Sometimes you need to keep intact the original behavior for code that relies on it, but you want to provide a fixed version moving forward for new code. You can do this by creating a new class that inherits from the original but overrides the broken method and provides a fixed version. Existing code can then continue to use the base class, complete with the bug that it may be working around, but new code can take advantage of the fixed version. As the old code is inspected and tested, it can be migrated to the new version as well.
例如，假设你有一段 Java 代码，如清单 5.9所示。这是一个完全人为的例子，用来演示通常所说的栅栏柱错误。如果你有一道 100 英尺长的栅栏，每 10 英尺一个栅栏柱，你需要多少个栅栏柱？如果你说 10 个，那你并不为过。许多人忘记了在栅栏的零标记处还有一个多余的栅栏柱。这是一个常见的错误，也称为差一错误。的实现fencepostHeights只是返回所请求的栅栏柱的高度。问题是，在 Java 中，高度数组是从零开始的。这意味着第一个元素的索引为 0，第二个元素的索引为 1，依此类推。问题是大多数人不会这样考虑栅栏。假设客户端代码询问用户他或她想要哪个栅栏柱的高度，然后将该条目传递给栅栏，如。客户端代码必须在发送之前通过减 1fence.getHeightOfPost(p)来调整 的值，否则结果将是用户期望的柱子高度之后的柱子高度。客户端代码必须解决这个问题。顺便说一句，很容易争辩说，差一错误不是一个错误，而是 Java 数组的工作方式。这是事实。这也是 C 和 C++ 数组的工作方式。错误不在代码中；它存在于类实现者和使用该类的开发人员之间的意图沟通错误中。实现该类的开发人员是基于合理的假设来实现的。使用该类的客户端也有合理的假设。它们只是不是同一套假设。p
For instance, assume you have a piece of Java code such as in Listing 5.9. This is a totally artificial example to demonstrate what is commonly called the fencepost error. If you have a fence that’s 100 feet long, with a fencepost every 10 feet, how many fenceposts are you going to need? If you said 10, you’re in good company. Many people forget that there’s an extra fencepost at the zero mark on the fence as well. It is a surprisingly common bug, also known as an off-by-one error. The implementation for fencepostHeights simply returns the height of the requested fencepost. The problem is, the array of heights is zero-based in Java. This means that the first element has an index of 0, the second element has an index of 1, and so on. The problem is that most people don’t think of fences in this way. Assume that client code asks the user which fencepost he or she wants the height of, and then passes the entry to a fence, as in fence.getHeightOfPost(p). The client code must adjust the value of p by subtracting 1 before sending it in, or the result will be the height of the post after the one the user expects. The client code must work around this problem. By the way, it’s easy to argue that an off-by-one error isn’t a bug, that it is how Java arrays work. That much is true. It is also how C and C++ arrays work. The bug isn’t in the code; it is in the miscommunication of intent between the class implementer and the developers who use the class. The developer who implemented the class did so with reasonable assumptions. The client using the class also has reasonable assumptions. They’re just not the same set of assumptions.
Listing 5.9. Implementation assumption mismatch.
public class Fence {
2 int [] fencepostHeights;
public int getHeightOfPost( int post ) {
4 return fencepostHeights[post];
};
6 };
8 ...
Fence fence;
10 // 栅栏在某个时刻被填满
...
12 Scanner scanner = new Scanner (System.in);
System.out.println(
14 "输入␣栅栏␣柱子␣以␣获取␣高度␣为：");
int p = scanner.nextInt ();
16 System.out.format("柱子␣ #%d ␣具有␣高度␣为：␣ %d%n",
p, fence.getHeightOfPost(p - 1));
18 // 错误解决方法 ^^^^
public class Fence {
2 int[] fencepostHeights;
public int getHeightOfPost( int post ) {
4 return fencepostHeights[post];
};
6 };
8 ...
Fence fence;
10 // Fence gets filled in at some point
...
12 Scanner scanner = new Scanner (System.in);
System.out.println(
14 "Enter␣fence␣post␣to␣get␣height␣of:");
int p = scanner.nextInt ();
16 System.out.format("Post␣#%d␣has␣height␣of:␣%d%n",
p, fence.getHeightOfPost(p - 1));
18 // Workaround for error ^^^^
从清单 5.9开始，当有人意识到不匹配并决定修复时会发生什么fencepostHeights（清单 5.10）？
Starting with Listing 5.9, what happens when someone realizes the mismatch and decides to fix fencepostHeights (Listing 5.10)?
Listing 5.10. Obvious fix—but likely not feasible.
public class Fence {
2 int [] fencepostHeights;
public int getHeightOfPost( int post ) {
4 return fencepostHeights[post - 1];
// 修复了 off-by-one 的错误 ^^^^
6 };
};
public class Fence {
2 int[] fencepostHeights;
public int getHeightOfPost( int post ) {
4 return fencepostHeights[post - 1];
// Bug fix for off-by-one ^^^^
6 };
};
Listing 5.11. Fixing a bug while leaving old code in place.
1 public class MendedFence扩展了Fence {
public int getHeightOfPost( int post ) {
3 return fencepostHeights[post - 1];
};
5 };
1 public class MendedFence extends Fence {
public int getHeightOfPost( int post ) {
3 return fencepostHeights[post - 1];
};
5 };
这个明显、直接且简单的修复只是破坏了调整发送到 的值的每一段客户端代码。现在客户端代码将使用预期高度左侧getHeightOfPost的帖子高度。这个修复可能没有预期的那么有用。如果有大量客户端，或者他们的时间表不同（几乎总是如此），那么几乎不可能协调所有团队同时纳入修复。这些情况下的通常结果是错误持续存在，并且必须将解决方法纳入每个客户端。这种解决方案很脆弱，容易出错。4
This obvious, straightforward, and simple fix just broke every piece of client code that was adjusting the value sent in to getHeightOfPost. Now the client code will be using the height of the post to the left of the expected one. This is probably not as helpful a fix as was intended. If there is a large number of clients, or if they are on different schedules—which is almost always the case—then it’s nearly impossible to coordinate all teams to incorporate the fix at once. The usual outcome in these cases is that the bug perpetuates, and the workaround must be incorporated into every client. This solution is fragile and prone to error.4
修复此问题的更好方法是提供该类的固定版本Fence，以供新客户使用，并允许旧客户按照自己的时间表迁移到该版本。修复方法如清单 5.11所示，基于清单 5.9Fence中的原始类。新开发的代码可以使用此版本。它更干净，不需要记住调整帖子数量。旧客户代码可以继续使用旧代码，直到有时间迁移到新类。一旦所有客户代码都使用新类，它和旧类就可以合并为一个类，供所有人使用。请注意，当源代码不可用时，此修复方法也可以工作。当您使用来自其他开发人员的库时，这种情况很常见。MendedFenceFence
A better way to fix this is to provide a fixed version of the Fence class that can be used by new clients and to allow old clients to migrate to it on their own schedule. The fix is shown in Listing 5.11, based on the original Fence class in Listing 5.9. Newly developed code can use this version instead. It is cleaner and doesn’t need to remember to adjust the post number. Old client code can continue using the old code until it has time to move to the new MendedFence class. Once all client code is using the new class, it and the old one can be merged into one class for everyone. Note that this fix can also work when the source code for Fence isn’t available. This is common when you’re using a library from another developer.
适用性
在以下情况下使用继承：
Use Inheritance when:
• 现有类提供的接口、实现和数据存储几乎满足新类的需求，但不完全满足。
• An existing class provides an interface, implementation, and data storage that is almost, but not quite, what is needed for a new class.
• 从原始类复制并粘贴代码要么出于维护原因而不合需要，要么由于原始源代码不可用而无法实现。
• Copying and pasting the code from the original class is either undesirable for maintenance reasons or not possible because the original source code is unavailable.
结构
參與者
超类
Superclass
系统中现有的类，用作产生进一步类的基础。
An existing class in a system that is used as a basis for producing a further class.
子类
Subclass
依赖于第一个类来实现其基本接口和实现的次要类。
The secondary class that relies on the first for its basic interface and implementation.
合作
该类Superclass创建一组基本方法接口以及任何附带的可选方法实现。Subclass继承了的所有接口元素Superclass，并且默认情况下还继承了所有实现，它可以选择用新的实现覆盖它。
The Superclass class creates a basic set of method interfaces with any accompanying optional method implementations. Subclass inherits all the interface elements of Superclass, and, by default, all of the implementations as well, which it may choose to override with new implementations.
结果
继承是一种强大的机制，但它也有一些有趣的限制和后果。首先，由于继承的核心有点微妙，面向对象编程理论，子类不能删除方法或数据字段。但可以非正式地说，继承在子类和超类之间创建了所谓的 IsA 关系。子类实际上是“超类的”专门版本。接口构成了类的定义，因此通过定义接口来说明超类是什么，然后在子类中转身删除该接口的部分内容，意味着子类不再是超类的一切。是的，这一切都相当存在主义，但这就是面向对象理论。最好坚持更简单的概念：如果不能毫无保留地将子类描述为超类的实例，则不要进行子类化。
Inheritance is a powerful mechanism that has some interesting limitations and consequences. For one thing, due to a bit of subtlety at the heart of object-oriented programming theory, a subclass may not remove a method or data field. It can be stated informally, however, that inheritance creates what is called an IsA relationship between the subclass and the superclass. The subclass literally “Is A” specialized version of the superclass. The interface is what constitutes the definition of what the class is, so saying what the superclass is by defining an interface, and then turning around and removing portions of that interface in a subclass, means that the subclass no longer is everything the superclass is. Yes, it’s all rather existential, but that’s object-oriented theory for you. Best to stick with the simpler concept: If the subclass cannot be described without reservation as an instance of the superclass, then do not subclass.
许多语言提供多重继承，允许子类具有多个超类。这实际上允许将子类描述为每个超类。例如，牛奶既是液体又是食物。如果一个类Milk要从继承Fluid，那么它可以用属性来描述Fluid，例如粘度和凝固点。如果Milk还从继承FoodItem，那么您可以讨论其营养成分或保质期。然而，在这种情况下需要考虑一些问题，许多语言拒绝多重继承，而倾向于更简单的单继承模型。
Many languages offer multiple inheritance, allowing a subclass to have multiple superclasses. This effectively allows a subclass to be described as each of the superclasses in turn. For instance, milk is both a fluid and a foodstuff. If a class Milk were to inherit from Fluid, then it could be described by the properties of a Fluid, such as viscosity and freezing point. If Milk also inherited from FoodItem, then you could discuss its nutritional content or expiration date. There are issues to be considered in such situations, however, and many languages reject multiple inheritance in favor of the simpler single-inheritance model.
在某些语言中，子类可以通过将方法设为私有来隐藏该方法，但这并不通用。在所有语言中，子类都可以覆盖方法并仅提供空实现，从而有效地消除行为，同时接口保持不变。
In some languages, a subclass may hide a method by making it private, but this is not universal. In all languages, a subclass may override a method and simply provide an empty implementation, effectively erasing the behavior while the interface remains the same.
重写似乎浪费了基类中的好代码，而且在很多情况下确实如此。扩展方法模式通过重写方法同时仍使用它来解决此问题。
It may seem that overriding is a waste of good code in the base class, and this is true in many cases. The Extend Method pattern solves this problem by overriding a method while still using it.
在某些情况下，您可能不希望继承整个现有类及其接口，而只想继承一小部分功能。如果原始源代码不可用，您可能对方法体的实现缺乏信心，当只需要该类的一小部分时，您可能不愿意承担将大型类集成到当前系统的成本，或者您可能有各种其他原因只使用现有类的一部分。
In some cases, you may not wish to inherit an entire existing class and its interface but instead to inherit only small pieces of functionality. You may lack confidence in the implementations of the method bodies if the original source code is unavailable, you may be reluctant to absorb the cost of integrating a large class into the current system when only a small segment of that class is needed, or you may have a variety of other reasons for using only part of the existing class.
当你希望保留部分（而非全部）接口时，请考虑使用重定向模式。使用Create Object将希望重用的类的实例放置在新类中，并将接口的一部分您想要保留的方法被复制，如清单 5.12所示。在这里，Square类实际上已经删除了setWidth:andHeight之前必须解决的方法。缺点是您必须为想要保留的每种方法实现包装器，一直追溯到继承层次结构。例如，清单 5.12包装了Redirection实例Shape中的方法。5您实际上是在删除原始类中未使用的方法，但是这样做时您无法利用静态类型语言中的多态性。动态类型语言，特别是那些允许在运行时确定方法可用性的语言，如 Objective-C 和 JavaScript，允许您更宽松地使用类型，在不存在子类关系的类型之间具有有效的多态性。基于原型的语言，如 JavaScript、Self 和 Lua，几乎普遍提供动态类型，并允许临时多态性而不受惩罚。
Consider using the Redirection pattern when you wish to retain some, but not all, of an interface. An instance of the class you wish to reuse is placed in the new class using Create Object, and the portion of the interface you want to retain is replicated, as in Listing 5.12. Here, the Square class has effectively removed the setWidth:andHeight method that had to be worked around before. The downside is that you have to implement wrappers for every method you wish to retain, all the way back up the inheritance hierarchy. For instance, Listing 5.12 wraps methods from Shape in Redirection instances.5 You are effectively erasing the unused methods of the original class, but you do not get to take advantage of polymorphism in statically typed languages when you do so. Dynamically typed languages, particularly those that allow for runtime determination of method availability, such as Objective-C and JavaScript, let you play with the typing a bit more loosely, having effective polymorphism among types not in a subclassing relationship. Languages based on prototyping, such as JavaScript, Self, and Lua, offer dynamic typing almost universally and allow ad hoc polymorphism without penalty.
Listing 5.12. Using Redirection to hide part of an interface.
1 @interface Square
{
3 矩形 rect;
}
5 - ( void ) setPosWithX: ( int ) x andY: ( int ) y;
- ( void ) setColor: (Color) c;
7 - ( void ) setLineWidth: ( int ) lw;
- ( void ) setSize: ( int ) s;
9 @end
11 @implementation Square
- ( void ) setPosWithX: ( int ) x andY: ( int ) y
13 {
[rect setPosWithX: x andY: y];
15 }
- ( void ) setColor: (Color) c
17 {
[rect setColor: c];
19 }
- ( void ) setLineWidth: ( int ) lw
21 {
[rect setLineWidth: lw];
23 }
- ( void ) setSize: ( int ) s
25 {
[rect setWidth: s andHeight: s];
27 }
@end
1 @interface Square
{
3 Rectangle rect;
}
5 - (void) setPosWithX: (int) x andY: (int) y;
- (void) setColor: (Color) c;
7 - (void) setLineWidth: (int) lw;
- (void) setSize: (int) s;
9 @end
11 @implementation Square
- (void) setPosWithX: (int) x andY: (int) y
13 {
[rect setPosWithX: x andY: y];
15 }
- (void) setColor: (Color) c
17 {
[rect setColor: c];
19 }
- (void) setLineWidth: (int) lw
21 {
[rect setLineWidth: lw];
23 }
- (void) setSize: (int) s
25 {
[rect setWidth: s andHeight: s];
27 }
@end
将这种情况与以下情况进行比较：您想要重用类的实现和/或数据存储，但对接口不满意并希望提供一个新的接口。在这种情况下，您可以在新类中使用对象实例，就像在重定向示例中所做的那样，但通过委托EDP 访问它。现在，您正在删除原始接口并提供一个新接口，同时仍然使用底层数据存储和行为实现。此功能在用作外观的类中很常见，这些类可在接口或 API 之间进行适配。
Compare this situation with one in which you want to reuse the implementation and/or data storage of a class but are unsatisfied with the interface and wish to provide a new one. In this situation, you can use an object instance in the new class, much like was done in the Redirection example but access it via the Delegation EDP. Now you are erasing the original interface and providing a new one while still using the underlying data storage and implementation of behaviors. This feature is common in classes that are to be used as facades, those that adapt between interfaces or APIs.
这些使用重定向和委托组合来解决继承限制的技术非常常见，以至于某些语言为该功能提供本机支持，例如 C# 的关键字。delegate
These techniques to work around limitations of Inheritance using a combination of Redirection and Delegation are so common that some languages provide native support for the feature, such as with C#’s delegate keyword.
执行
创建继承关系的机制因语言而异，但静态类型语言几乎总是提供清晰的语法。动态类型或基于原型的语言可能并不总是使关系明确。
The mechanism for creating an inheritance relationship varies from language to language, but statically typed languages almost always provide clear syntax for doing so. Dynamically typed or prototype-based languages may not always make the relationship explicit.
在 C++ 中：
In C++:
类超类 {
2 public:
超类() ;
4 };
6类 子类:公共超类 {
public:
8 子类() ;
};
class Superclass {
2 public:
Superclass( );
4 };
6 class Subclass : public Superclass {
public:
8 Subclass( );
};
在 Python 中：
In Python:
1 类超类（对象）：
def __init__(self):
3 通过
5 类子类（超类）：
def __init__(self):
7 超类.__init__(self)
1 class Superclass(object):
def __init__(self):
3 pass
5 class Subclass(Superclass):
def __init__(self):
7 Superclass.__init__(self)
相关模式
继承无处不在。它是任何使用子类型或兄弟对象类型相似性的 EDP 的核心组件。具体而言，请参阅Trusted委托、代理委托、受信任重定向、代理重定向、恢复方法和扩展方法。此外，继承还用于Fulfill 方法模式。由于这些模式中的每一个都是许多其他组合模式的基础，因此继承是您应该彻底理解的 EDP。
Inheritance is ubiquitously used. It is a core component of any EDP that uses object type similarities of Subtype or Sibling. In particular, see Trusted Delegation, Deputized Delegation, Trusted Redirection, Deputized Redirection, Revert Method, and Extend Method. In addition, Inheritance is used in the Fulfill Method pattern. Because each of these patterns is fundamental to many other composed patterns, Inheritance is an EDP you should make sure you understand thoroughly.
另请参阅委托和重定向以获取有关如何使用这些模式来解决继承的某些限制的更多信息。
Also see Delegation and Redirection for further information on how to use those patterns to work around some limitations of Inheritance.
抽象接口
Abstract Interface
类型关系
Type Relation
意图
提供通用接口，用于在对象类型系列中应用行为，但不提供实际操作的实现。在这种情况下，由于不存在默认方法，子类被迫提供自己的适当实现。
To provide a common interface for applying a behavior in a family of object types but without providing an implementation of the actual operation. In this scenario, subclasses are forced to provide a proper implementation of their own because no default method exists.
也称为
虚方法、多态、推迟实现
Virtual Method, Polymorphism, Defer Implementation
动机
通常，当我们使用符合我们概念设计的继承模式来构建类层次结构时，我们会遇到一种情况，即我们根本无法在类层次结构的根部提供有意义的方法实现。我们知道我们想要在概念上做什么，但我们并不完全确定如何去做。
Often, when we have a hierarchy of classes using the Inheritance pattern that conforms to our conceptual design, we run into a situation where we simply cannot provide a meaningful method implementation at the root of the class hierarchy. We know what we want to do conceptually, but we are not entirely sure how to go about doing it.
假设我们正在模拟动物。粗略地说，所有动物都有特定的行为和需求。它们进食、衰老，并且（通常）会移动。它们吃什么取决于物种，但进食几乎总是某种摄取行为。有些动物实际上可能不摄取食物，但它们是罕见的特殊情况。动物在成熟过程中会经历多个阶段，例如妊娠期、幼年期和成年期。每个阶段的细节和时间因物种而异，但至少可以模拟一个广泛的默认衰老过程。
Assume that we are modeling animals. Animals all, roughly speaking, have certain behaviors and needs. They eat, they age, and they (usually) move. What they eat depends on the species, but eating is almost always an act of ingestion of some sort. Some animals may not actually ingest food, but they are rare and special cases. Animals pass through a number of phases as they mature, such as gestation, juvenile, and adult. The details and timing of each stage vary across species, but at least a broad default process of aging can be modeled.
然而，移动则有点棘手。我们可以为动物提供一个移动的目标位置，但实现这种行为是一项更复杂的任务：鱼类、鸟类和陆生动物都有非常不同的运动方式和它们可以到达的位置。我们在这里无法真正模拟过于宽泛的行为，至少在不需要从一开始就立即使用大量moveTo特殊情况的情况下无法实现。无论我们选择如何实现方法，我们最终都会在大多数子类中或多或少地完全覆盖和替换实现。我们最终可能会得到如清单 5.13中所示的结果。
Movement, however, is a bit trickier. We can give a destination location for the animal to move to, but implementing this behavior is a more complex task: fish, birds, and terrestrial animals all have very different modes of locomotion and kinds of locations that they can get to. There’s no overly broad behavior that we can really model here, at least not without requiring a lot of special cases immediately from the beginning. However we choose to implement a moveTo method, we’re going to end up overriding and replacing the implementation more or less completely in most of the subclasses. We’re likely going to end up with something such as in Listing 5.13.
这告诉我们，也许我们根本不应该提供实施方案，因为无论我们选择什么，它几乎肯定是错误的。我们希望确保每个Animal，无论哪个子类描述它，都可以被要求移动，但我们无法提出适用于所有甚至大多数动物的合理默认移动方式。我们希望提供一个接口— 一种调用行为的定义方式 — 但没有定义行为本身的实现。
This tells us that perhaps we shouldn’t offer an implementation at all, because no matter what we choose, it’s almost certainly going to be wrong. We want to ensure that every Animal, no matter which subclass describes it, can be asked to move, but we can’t come up with a reasonable default for how to move that would apply to all or even most animals. We want to provide an interface—a defined way of invoking the behavior—but no implementation to define the behavior itself.
清单 5.13. 几乎所有动物都会移动，但是方式却大不相同。
Listing 5.13. Animals almost all move but in very different ways.
1 public class Animal {
public void eatFood( FoodItem f ) {
3 this .ingest(f);
this .digest(f);
5 };
public void matureTo ( TimeDuration age ) {
7 if (age > gestation) {
this .beAJuvenile(age);
9 } else if (age > maturation) {
this .beAnAdult(age);
11 } else if (age > longevity) {
this .die();
13 }
};
15 public void moveTo( Location dest ) {
// 嗯...这里该放什么？
17 };
};
1 public class Animal {
public void eatFood( FoodItem f ) {
3 this.ingest(f);
this.digest(f);
5 };
public void matureTo ( TimeDuration age ) {
7 if (age > gestation) {
this.beAJuvenile(age);
9 } else if (age > maturation) {
this.beAnAdult(age);
11 } else if (age > longevity) {
this.die();
13 }
};
15 public void moveTo( Location dest ) {
// Um... what to put here?
17 };
};
回想一下继承中的讨论，类型中的行为由类型上定义的方法提供。因此，为了在所有子类中均匀地提供行为，我们应该提供方法定义，但我们却陷入困境，因为我们没有定义方法的实现。
Recall from the discussion in Inheritance that behaviors in types are provided by the methods that are defined on a type. Therefore, to provide the behavior evenly across all the subclasses, we should provide a method definition, but we’re then stuck because we have no implementation to define the method with.
如果我们提供一个空的实现主体（如清单 5.13所示） ，那么我们就提供了一个什么都不做的默认实现。如果子类未能重写此方法，那么该类所模拟的可怜动物将根本无法移动。虽然有些动物确实静止不动，但它们是例外。因此，在这种情况下，我们还希望确保提醒子类的实现者定义适当的实现，即使我们做不到。解决方案是为该方法创建一个抽象接口。
If we provide an empty implementation body as in Listing 5.13, then we’ve provided a default implementation that simply does nothing. If a subclass fails to override this method, then the poor animal being modeled by that class won’t be able to move at all. Although there are some truly stationary animals, they are the exceptions. Therefore, in cases such as this, we would also like to ensure that the implementors of subclasses are reminded to define a proper implementation, even though we can’t. The solution is to create an Abstract Interface for that method.
适用性
在以下情况下使用抽象接口：
Use Abstract Interface when:
• 在类定义时不知道方法的实现，或者无法确定合理的默认实现。
• Implementation of a method either is not known at class definition time or no reasonable default implementation can be determined.
• The interface for that method can be determined.
• 您希望子类根据其特殊需要来处理该方法的功能。
• You expect subclasses to handle the functionality of the method according to their special needs.
结构
參與者
摘要者
Abstractor
声明操作接口的类类型。
The class type that declares an interface for operation.
手术
operation
被抽象出来的方法：无法给出方法定义。
The method being abstracted out: no method definition can be given.
合作
抽象类为所有子类都将实现的方法定义一个接口。（Abstractor此处与尚未知晓的子类之间存在隐式协作。）
Abstractor defines an interface for a method that all subclasses will implement. (An implicit collaboration occurs between Abstractor and an as-yet unknown subclass here.)
结果
创建对象模式允许我们实例化对象，而检索模式则显示如何填充新创建对象的字段。抽象接口的不同寻常之处在于它表示缺少方法实现；它不是向我们展示如何填充方法，而是向我们展示如何将方法定义推迟到设计中的另一个点。有关解决方案的其余部分，请参阅相关的实现方法模式。
The Create Object pattern lets us instantiate objects, and the Retrieve pattern shows how to fill in the fields of that newly created object. Abstract Interface is unusual in that it indicates the absence of a method implementation; instead of showing us how to fill in the method, it shows us how to defer the method definition until another point in the design. See the related Fulfill Method pattern for the rest of the solution.
在这种情况下，声明方法是为了定义适合我们概念需求的接口，但方法主体未定义。这并不意味着我们只是定义了一个空方法，一个不执行任何操作的方法；该方法根本没有定义。这是一个关键点，也是面向对象语言新手程序员经常忽略的一点。如何做到这一点因语言而异，正如您在实现部分中的示例中所看到的。
In this case, the method is declared in order to define the proper interface for our conceptual needs, yet the method body is left undefined. This does not mean that we simply define an empty method, one that does nothing; the method has no definition at all. This is a critical point, and one that is often missed by programmers new to object-oriented languages. How it is done varies from language to language, as you can see from the examples in the Implementation section.
不同语言对这种模式的使用也不同。例如，C++ 断言，任何类类型，即使只有一个抽象接口方法，也是抽象类，因此无法直接实例化。只有通过Fulfill Method模式为抽象方法提供定义的子类，才可以使用Create Object实例化为对象。
Different languages have different uses of this pattern as well. C++, for instance, asserts that any class type with even a single method that is an Abstract Interface is an abstract class and therefore incapable of being instantiated directly. Only subclasses that provide a definition for the abstract method via the Fulfill Method pattern may be instantiated into objects using Create Object.
Java 则更进一步。与 C++ 一样，类可能包含一个或多个抽象方法，从而使类成为抽象类。但是，如果它只包含抽象方法，Java 会提供接口构造。这里的术语接口是指收集许多抽象方法的类状实体，而不是单个方法的接口。与抽象类一样，Java 接口无法实例化。与可以同时具有已定义和抽象方法的抽象类不同，Java 接口根本不允许具有任何已定义方法。类应该至少有一个已定义的方法，而接口仅由抽象方法组成。尽管 Java 只有类的单一继承，但它提供接口的多重继承，以允许以非常细的粒度组合新的类接口。有关继承模型的更多信息，请参阅继承。
Java takes it even further. As with C++, a class may contain one or more abstract methods, making the class abstract. If it contains nothing but abstract methods, however, Java offers the interface construct instead. The term interface here refers to a classlike entity collecting many abstract methods, not the interface of a single method. Like an abstract class, a Java interface cannot be instantiated. Unlike abstract classes, which can have both defined and abstract methods, a Java interface is not allowed to have any defined methods at all. Classes should have at least one defined method, and interfaces are composed of only abstract methods. Although Java has only single inheritance of classes, it offers multiple inheritance of interfaces to allow the composition of new class interfaces at a very fine granularity. See Inheritance for more information on inheritance models.
Python 仅在库的 2.6 版本中增加了对抽象方法的支持，作为abc模块中的标准库附加项，模块代表抽象基类。Python 3.0 引入了一种稍微更直接的符号，您将在实现部分看到它。与 Java 或 C++ 不同，Python 的方法允许默认实现，子类可以调用它，通常使用扩展方法EDP。但是，子类仍然必须覆盖该方法并提供自己的实现。
Python added support for abstract methods only in version 2.6 of the library as a standard library addition in the abc module, which stands for abstract base class. Python 3.0 introduced a slightly more straightforward notation, which you will see in the Implementation section. Unlike Java or C++, Python’s approach allows for a default implementation, which can be called by subclasses, most often using the Extend Method EDP. Subclasses, however, still must override the method and provide their own implementation.
执行
在 C++ 中，该方法被设置为零并称为纯虚方法：
In C++ the method is set equal to zero and called a pure virtual method:
类AbstractOperations {
2 公共：
虚拟 void operation() = 0;
4 };
6 类DefinedOperations：
公共AbstractOperations {
8 公共：
void operation();
10 };
12 void
DefinedOperations::operation() {
14 // 执行适当的工作
};
class AbstractOperations {
2 public:
virtual void operation() = 0;
4 };
6 class DefinedOperations :
public AbstractOperations {
8 public:
void operation();
10 };
12 void
DefinedOperations::operation() {
14 // Perform the appropriate work
};
在 Java 中，该方法包含在接口中或位于抽象类中：
In Java the method is included in an interface or is in an abstract class:
1 public interface AbstractOperations {
public void operation();
3 };
5 public abstract class SemiDefinedOperations {
public abstract void operation2();
7 public void operation3() {};
};
9
public class DefinedOperations
11 extends SemiDefinedOperations
implements AbstractOperations {
13 public void operation() {
// 执行适当的工作
15 }
public void operation2() {
17 // 执行适当的工作
}
19 };
1 public interface AbstractOperations {
public void operation();
3 };
5 public abstract class SemiDefinedOperations {
public abstract void operation2();
7 public void operation3() {};
};
9
public class DefinedOperations
11 extends SemiDefinedOperations
implements AbstractOperations {
13 public void operation() {
// Perform the appropriate work
15 }
public void operation2() {
17 // Perform the appropriate work
}
19 };
在 Python 3.x 中：
In Python 3.x:
1 class AbstractOperations(metaclass=ABCMeta):
@abstractmethod
3 def operation(self, ...):
// 允许默认实现
5 return
7 class DefinedOperations(AbstractOperations):
def operation(self, ...):
9 // 执行适当的工作
传递
1 class AbstractOperations(metaclass=ABCMeta):
@abstractmethod
3 def operation(self, ...):
// Default implementation allowed
5 return
7 class DefinedOperations(AbstractOperations):
def operation(self, ...):
9 // Perform the appropriate work
pass
相关模式
显然，继承将与此 EDP 结合使用，以设置子类，该子类将为抽象接口中使用的方法提供实现。有关如何执行此操作的完整信息，请参阅第 6 章中的Fulfill Method模式。
Inheritance will obviously be used in conjunction with this EDP to set up the subclass that will provide an implementation for the method used in Abstract Interface. See the Fulfill Method pattern in Chapter 6 for the complete story on how to go about doing so.
代表团
Delegation
对象行为
Object Behavioral
意图
将当前工作的一部分分发或委托给另一个对象的另一个方法。
To parcel out, or delegate, a portion of the current work to another method in another object.
也称为
消息传递、方法调用、调用、执行
Messaging, Method Invocation, Calls, The Executive
动机
在处理对象的过程中，经常会出现“其他某个对象”可以提供我们想要的某项功能的情况。委托体现了从一个对象到另一个对象的方法调用的最常见形式，允许一个对象向另一个对象发送消息，以执行一些工作。接收对象可能会也可能不会因此发回数据。
In the course of working with objects, the situation often arises that “some other object” can provide a piece of functionality we want to have. Delegation embodies the most general form of a method call from one object to another, allowing one object to send a message to another, to perform some bit of work. The receiving object may or may not send back data as a result.
举一个真实世界的例子，让我们来看一下一家公司的运作方式。首席执行官的目标是成功经营一家公司。她指派下属负责公司的不同部分，而且很少有两个下属负责同一部分运营。例如，财务副总裁确保财务报告符合政府标准，首席技术官确保满足公司的技术需求，等等。每项工作都是独立的，与其他工作以及首席执行官的职责都有很大不同，但整个公司的成功依赖于所有部分的协同作用。清单 5.14展示了一些模拟这种方法的代码。
As a real-world example, consider how a corporation works. The CEO’s goal is to successfully run a company. She assigns subordinates to handle different parts of it, and rarely are any two subordinates in charge of the same part of the operation. The vice president of finance, for instance, ensures that the financial reporting meets government standards, the CTO makes sure the technology needs of the company are addressed, and so on. Each job is discrete and differs greatly from the others as well as from the responsibilities of the CEO, but the success of the company as a whole relies on the synergy of all of its parts. Listing 5.14 shows a bit of code that models this approach.
Listing 5.14. CEO delegates out responsibilities.
公共类CEO {
2 FinanceExec vpFinance;
ResearchExec vpResearch;
4 TechnologyExec cto;
公共无效runCompany（）{
6 vpFinance.ensureFinancialCompliance（）;
vpResearch runResearchDivision（）;
8 cto.manageTechnology（）;
};
10 };
public class CEO {
2 FinanceExec vpFinance;
ResearchExec vpResearch;
4 TechnologyExec cto;
public void runCompany () {
6 vpFinance.ensureFinancialCompliance();
vpResearch runResearchDivision();
8 cto.manageTechnology();
};
10 };
这里值得注意的一点是，清单 5.14中的例程是同步方法调用的一个示例。在此 Java 代码中，当runCompany调用时，要求财务副总裁管理财务，并且只有当他完成报告后，才会要求研究副总裁开始他的任务。任务由调用任务同步。首先执行一个任务，然后执行另一个任务，并按特定顺序执行。
One interesting point to make here is that the routine in Listing 5.14 is an example of synchronous method calling. In this Java code, when runCompany is invoked, the vice president of finance is asked to manage the finances, and only when he is finished reporting is the vice president of research asked to start his task. The tasks are synchronized by the calling task. First one, then the other, and execution is done in a specific order.
当然，在现实生活中，首席执行官的工作方式并非如此。她要求所有高管同时去做自己的工作。然后，当每个人完成后，他或她会汇报自己的日程安排。这是异步执行的一个例子。代表们被告知要并行去做自己的事情。
In real life, of course, this isn’t how the CEO performs her job. She asks all the executives to go do their jobs and to do them at the same time. Then, when each is done, he or she reports back on his or her own schedule. This is an example of asynchronous execution. The delegates are told to go off and do their own thing, in parallel.
这种异步或并行处理在大多数面向对象语言中都不是一个简单的功能，尽管有些语言确实提供了线程库（如 Java 的FutureTaskAPI）来满足这一需求。其他语言，尤其是 Go 和 Erlang 等函数式语言，通过异步调用原生支持并发和并行功能，但正如您所预料的那样，它们的语法完全不同。无论它被称为异步调用、并发编程还是并行编程，它都是一门超出本书范围的学科。但请注意，本文中描述的每个 EDP 都适用于异步调用。
Such asynchronous, or parallel, processing is not a simple feature to perform in most object-oriented languages, although some do offer threading libraries, such as Java’s FutureTask API, to fulfill this need. Other languages, particularly functional languages such as Go and Erlang, have native support for concurrency and parallel features via asynchronous calls, but their syntax is, as you might expect, radically different. Whether it is referred to as asynchronous calling, concurrent programming, or parallel programming, it is an entire discipline beyond the scope of this book. Note, however, that every EDP described in this text is applicable as an asynchronous call.
委托是一种十分常见的设计和功能组件，C# 提供了该delegate关键字。此语言特性将函数封装在对象中，这样就可以像常规对象一样传递该函数。然后，此委托对象就可以像常规方法一样被调用。封闭对象本质上是不可见的。6这里的要点是，此功能不对被包装的方法做出任何断言，也不施加任何限制，只是要求它与参数和返回值的预定义类型集合相匹配。被调用的委托对象显然与调用对象不同，对象类型彼此无关，被包装的方法可以命名为执行调用的方法，也可以不命名为执行调用的方法。这确实是最普遍的情况，也是委托的一个例子。
Delegation is a common enough component of design and functionality that C# offers the delegate keyword. This language feature encapsulates a function in an object such that it can be passed around as if it were a regular object. This delegate object then becomes callable as a regular method would be. The enclosing object is essentially invisible.6 The point here is that this feature makes no assertions and imposes no restrictions on the method being wrapped other than that it match a predefined set of types for the arguments and return value. The delegate object being called upon is obviously dissimilar from the calling object, the object types have nothing to do with one another, and the method being wrapped may or may not be named anything like the method doing the calling. This truly is the most general possible case and an example of Delegation.
适用性
在以下情况下使用委派：
Use Delegation when:
• 另一个对象可以执行当前方法主体希望完成的一些工作。
• Another object can perform some work that your current method body wishes to have done.
• 另一个对象不需要访问当前对象中的私有数据即可完成任务。
• The other object does not need access to the private data in the current object to complete the task.
• 两个对象之间没有已知的相关类型关系。
• There is no known relevant type relationship between the two objects.
结构
參與者
委托人
Delegator
发送消息到 的对象类型Delegate。
The object type sending the message to the Delegate.
手术
operation
Delegator发送消息时当前正在执行的方法— — operation2方法调用的调用点。
The method within the Delegator that is currently being executed when the message is sent—the point of invocation of the operation2 method call.
代表
Delegate
接收消息的对象类型，并带有要调用的适当方法。
The object type receiving the message, with an appropriate method to be invoked.
操作2
operation2
从调用站点调用的方法。
The method being invoked from the call site.
合作
这是一个简单的二元关系：一个方法调用另一个方法，就像在过程系统中一样，并且具有相同的注意事项和要求。但是，因为我们在面向对象领域工作，所以我们有几个额外的需求。我们要求被调用的对象在调用时可见，无论是通过本地范围的对象变量，还是通过Retrieve访问。此外，显然，被调用的方法必须对其他对象可见。
This is a simple binary relationship: one method calls another, just as in procedural systems, and with the same sorts of caveats and requirements. Because we are working in an object-oriented realm, however, we have a couple of additional needs. We require the object being called to be visible at the point of invocation, whether by being a locally scoped object variable, or by access through Retrieve. Also, obviously, the method being called must be visible to other objects.
结果
任何两个对象之间的所有操作都可以描述为委托模式的一个实例，但能够描述关系的进一步属性更有用。请参阅其余方法调用 EDP，了解更适合特定任务和需求的委托细化。然而，这种方法调用的广义形式是一些高级抽象（如Bridge或其对象变体形式的Adapter ）中的关键概念，其目的是在两个原本不相关的接口之间创建格式良好且有效的转换。如果接口以有条不紊的方式相关，那么这些模式不一定是正确的选择。因此，接口和类型之间缺乏关系是一种必需的特性，而委托满足了这一需求。
All operations between any two objects can be described as an instance of the Delegation pattern, but being able to describe further attributes of the relationship is much more useful. See the remaining Method Invocation EDPs for refinements of Delegation that are better suited to specific tasks and needs. This generalized form of a method call, however, is a crucial concept in some of the higher abstractions such as Bridge, or Adapter in its object-variant form, where the point is to create a well-formed and effective translation between two interfaces that are otherwise unrelated. If the interfaces are related in a methodical manner, then these patterns are not necessarily the right ones to use. The lack of relation between the interfaces and types is a required trait, therefore, and Delegation fulfills that need.
执行
委托是面向对象编程中最通用、最基本的方法调用方式，它描述了两个对象作为消息的发送者和接收者如何相互通信、执行工作并返回值。
The most generalized and basic style of method invocation in object-oriented programming, Delegation describes how two objects communicate with each other, as the sender and receiver of messages, performing work and returning values.
在 C++ 中：
In C++:
class Delegator {
2 public:
Delegatee target;
4 void operation() {
// 工作可以在之前完成...
6 target.operation2();
// 工作可以在之后完成...
8 };
};
10
class Delegatee {
12 public:
void operation2();
14 };
class Delegator {
2 public:
Delegatee target;
4 void operation() {
// Work may be done before...
6 target.operation2();
// Work may be done after...
8 };
};
10
class Delegatee {
12 public:
void operation2();
14 };
相关模式
委托在面向对象编程中无处不在。如果没有其他方法调用 EDP 适用，则这是最通用的默认形式。直接相关的 EDP 包括通过对委托进行微小更改而达到的 EDP 。通过将方法相似性更改为相似，您将获得重定向。将对象类型相似性更改为相似会得到委托聚集。只修改对象相似性不会给我们任何特别有意义的东西，因为这会导致单个对象同时被施加两种不同的类型。委托通常与检索结合使用，后者提供要调用的对象。
Delegation is ubiquitous in object-oriented programming. If no other Method Invocation EDP applies, this is the most general form that will be the default. Immediately related EDPs include those reached through minor changes to Delegation. By altering the method similarity to similar, you arrive at Redirection. Changing the object type similarity to similar gives you Delegated Conglomeration. Modifying only the object similarity doesn’t give us anything particularly meaningful, because it would result in a single object having two dissimilar types imposed on it simultaneously. Delegation is often found in conjunction with Retrieve, which provides the object to be called upon.
方法调用分类
对象：不同
Object: Dissimilar
对象类型：不同
Object Type: Dissimilar
方法：不同
Method: Dissimilar
重定向
Redirection
对象行为
Object Behavioral
意图
请求另一个对象执行与当前任务紧密相关的子任务，也许是执行基本工作。
To request that another object perform a tightly related subtask to the task at hand, perhaps performing the basic work.
也称为
汤姆索亚，车间领班
Tom Sawyer, Shop Foreman
动机
重定向是对委托的一个小改进，它考虑到执行类似任务的方法通常具有相似的名称。我们可以利用这一点来阐明此模式相对于更通用形式的目的，并展示其适用的案例。
A small refinement to Delegation, Redirection takes into consideration that methods performing similar tasks are often named similarly. We can take advantage of this to clarify the intent of this pattern over the more general form and show cases where it is applicable.
汤姆索亚就是这种模式的一个例子，他是文学人物，因说服别人为他完成任务而闻名[ 39 ]。与代表之类的高管别名不同，汤姆通常希望别人完全按照他要求的方式完成任务。一个常见的轶事是，汤姆被要求粉刷他姑妈的篱笆。他不想亲自做这项工作，所以他说服了一群朋友帮他做。他本可以让一个朋友负责油漆桶，另一个朋友准备篱笆，等等，这样的工作流程看起来就会很像高管的工作流程。相反，他认为最好的分工方式是让每个朋友都执行同一项任务：粉刷篱笆。汤姆确保每个人都有合适的工具和足够的油漆，并且任务完成，但是他自己并不执行这项任务。他要求那些粉刷篱笆的人都做汤姆要求他们做的同一项任务——他们只是分了更小的部分。
An example for this pattern is Tom Sawyer, a literary figure who was famous for convincing other people to do his tasks for him [39]. Unlike the Executive alias of Delegation, Tom generally sought to have others do exactly the task he was asked to do. The common anecdote involves Tom being asked to paint his aunt’s fence. He didn’t want to do the work himself, so he convinced a group of friends to do it for him. He could have had one friend handle paint buckets, another prep the fence, and so on, and the workflow would have looked a lot like that of an executive. Instead, he decided the best way of dividing the work was to have each friend perform the same task: paint the fence. Tom made sure that each person had the right tools and enough paint and that the task was completed, but he did not perform the task himself. Those he asked to paint the fence each did the same task Tom was asked to do—they just did smaller portions of it.
重定向也是如此。当一项工作可以分解为具有与主要任务相同的基本动机和意图的较小子任务时，请使用重定向。请注意，在不同方法中行为可能截然不同，但意图是相同的。汤姆没有油漆——他分发油漆和刷子——但他的意图是粉刷篱笆。同样，为他工作的人也打算粉刷篱笆——他们只关心篱笆的较小部分。清单 5.15显示了一个例子Java 语言中：当汤姆接到要求时，他会指示他的每个朋友粉刷栅栏。
So it is with Redirection. Use Redirection when a job can be broken down into smaller subtasks that have the same basic motivation and intent as the main task. Note that the behavior may be radically different across the methods but that intent is the same. Tom didn’t paint—he handed out paint and brushes—but his intent was to get the fence painted. Likewise, the people working for him intended to paint the fence—they were just concerned with smaller sections of it. Listing 5.15 shows an example in Java: Tom directs each of his friends to paint the fence when he’s asked to do it.
Listing 5.15. Tom paints the fence with help.
public class Friend {
2 public void paintTheFence( int beg, int end) {
// 进行绘制
4 };
};
6
public class TomSawyer {
8 // Friends 在其他地方收集
java.util.ArrayList<Friend> friends;
10
// 从 beg 到 end 绘制栅栏
12 public void paintTheFence( int beg, int end) {
int fenceLength = end - beg;
14 int subfence = fenceLength / friends.size();
int friendBeg = beg;
16 for (Friend f : friends) {
int friendEnd = friendBeg + subfence - 1;
18 f.paintTheFence(friendBeg, friendEnd);
// ^--- 重定向
20 friendBeg = friendEnd + 1;
}
22 };
};
public class Friend {
2 public void paintTheFence(int beg, int end) {
// Do the painting
4 };
};
6
public class TomSawyer {
8 // Friends are collected elsewhere
java.util.ArrayList<Friend> friends;
10
// Paint the fence from beg to end
12 public void paintTheFence(int beg, int end) {
int fenceLength = end - beg;
14 int subfence = fenceLength / friends.size();
int friendBeg = beg;
16 for (Friend f : friends) {
int friendEnd = friendBeg + subfence - 1;
18 f.paintTheFence(friendBeg, friendEnd);
// ^--- Redirect
20 friendBeg = friendEnd + 1;
}
22 };
};
这种设置与我们在第 2 章第2.2.4 节中看到的车间领班喷漆汽车的情况类似。请注意，在这两种情况下，调用方法实际上都没有执行所请求的工作，尽管这当然是可能的。调用方法是否纯粹充当路由器、领班或经理，或者它是否参与实际执行任务并不重要。
This setup is similar to what we saw back in Chapter 2, Section 2.2.4, with the shop foreman painting cars. Notice that in neither case is the calling method actually doing the work requested, although it certainly is possible. Whether the calling method is acting purely as a router, a foreman, or a manager, or whether it is involved in actually performing the task is immaterial.
在其他时候，上门服务的人可能对要完成的任务有稍微不同的想法，并要求其他人完成大部分工作，然后上门服务的人要么进行准备工作，要么进行事后的清理。如果你要粉刷房间，有两种方法可以做到。你可以直接开始用刷子粉刷——希望不要把太多的油漆涂到电灯开关、装饰条、插座等东西上——或者你可以花时间仔细准备房间，铺上防水布并用胶带正确地粘在踢脚板上，取下电灯开关和插座板，用胶带粘住所有装饰条边缘，然后小心地把所有东西放回原处，然后取下胶带和防水布。
At other times, the caller may have a slightly different idea of what the task to be completed is and ask someone else to do the bulk of the work, and the caller will then either perform prep work or do cleanup afterwards. If you’re painting a room, there are two ways you can go about doing it. You can either dive in and start painting with a brush—and hope you don’t slop too much paint on lightswitches, molding, outlets, and the like—or you can take time to carefully prepare the room by laying down tarps properly taped to the baseboards, removing the lightswitch and outlet plates, taping all the molding edges, and carefully replacing everything and removing the tape and tarps afterwards.
一个能快速、熟练地涂刷一层油漆的人不一定是一个在准备工作和清理工作上细心、有条不紊的人。他们都在粉刷房间，但一个人的如何做到这一点的理念是最大限度地提高速度，而另一个人的重点是质量。如果可以的话，最好同时做这两件事。在这种情况下，你可以让细心的人进行准备工作和清理，然后这个人可以请别人来做实际的绘画工作，以加快速度。现在，准备和清理的人可以在第一个房间被粉刷的时候自由地去准备另一个房间。这个过程可以同步建模为类似清单 5.16 的内容。
Someone who is lightning fast and good at applying a coat of paint isn’t necessarily going to be the person who is careful and methodical about prep work and cleanup. They’re both painting the room, but one person’s concept of how to go about doing it is to maximize speed, while the other person’s focus is on quality. It would be great to do both where you can. In cases such as this, you can have the careful person perform the prep work and cleanup, and then that person can ask someone else to do the actual painting for speed. Now the prep and cleanup person is free to go prep another room while the first one is being painted. This process can be modeled synchronously as something like Listing 5.16.
Listing 5.16. Prep work and cleanup are important.
1 class SloppyFastPainter {
public:
3 void paintRoom( Room r ) {
// 粉刷墙壁
5 };
};
7
class CarefulPainter {
9 SloppyFastPainter sloppy;
public:
11 void paintRoom( Room r ) {
// 在房间中铺上防水布
13 // 从房间中移除硬件
sloppy.paintRoom( r ); // 重定向
15 // 更换房间中的硬件
// 清理房间
17 };
};
1 class SloppyFastPainter {
public:
3 void paintRoom( Room r ) {
// Paint the walls
5 };
};
7
class CarefulPainter {
9 SloppyFastPainter sloppy;
public:
11 void paintRoom( Room r ) {
// Laydown tarps in room
13 // Remove hardware from room
sloppy.paintRoom( r ); // Redirect
15 // Replace hardware in room
// Clean up room
17 };
};
不幸的是，这里很容易引起混淆。在一般意义上，委托这个词只是意味着将部分工作量交给其他人。委托EDP 定义的就是这个最普遍的意义。每个方法调用 EDP 都是委托的一种特殊化，但重定向最容易被误认为是委托。这种混淆是可以理解的，因为委托这个术语在软件工程中有多种使用方式。只需记住，即使是 C# 也已将委托这个术语用于最普遍的情况，使用它作为一个关键字来包装要传递的方法，同时保留我们在本文中定义的所有三个相似性轴。委托用于交给任何其他方法，而使用重定向则更具体一些。
Unfortunately, there is an opportunity for confusion here. Used in the general sense, the word delegate simply means to hand off part of a workload to someone else. That most general sense is what the Delegation EDP defines. Every method-call EDP is a specialization of delegation, but Redirection is the most likely one to be mistakenly described as Delegation. This confusion is quite understandable, as the term delegation is used in a number of ways in software engineering. Just remember that even C# has adopted the term delegate for the most general case, using it as a keyword wrapping a method to be passed around while leaving open all three of the similarity axes we defined in this text. Delegation is used for handing off to any other method, while using Redirection is a bit more specific.
适用性
在以下情况下使用重定向：
Use Redirection when:
• 另一个对象可以执行当前方法主体希望完成的一些工作。
• Another object can perform some work that your current method body wishes to have done.
• The other object does not need access to the private data in the current object to complete the task.
• 两个对象之间没有已知的相关类型关系。
• There is no known relevant type relationship between the two objects.
• 该目标对象拥有一个具有类似意图的方法，通过其签名名称来表达，该名称遵循 Kent Beck 的意图显示选择器模式。
• That target object has a method that has a similar intent, expressed through its signature name after Kent Beck’s Intention Revealing Selector pattern.
结构
參與者
重定向器
Redirector
方法调用的发起站点包含一个名为操作的方法，该方法有一个子任务要分配给另一个对象。此对象redirectTarget是 类型的字段元素Redirectee。
The originating site of the method call contains a method named operation, which has a subtask to be parceled out to another object. This object, redirectTarget, is a field element of type Redirectee.
重定向对象
Redirectee
执行所要求的子任务的消息接收者的类型。
The type of the receiver of the message, which performs the subtask asked of it.
手术
operation
重定向器的操作调用redirectTarget. 操作来执行其部分工作，从而调用操作Redirectee的版本。
Redirector’s operation calls redirectTarget. operation to perform a portion of its work, and thereby invokes Redirectee’s version of operation.
合作
与委托（以及所有方法调用 EDP）一样， Retrieve描述了两个对象及其所包含的方法之间的二元关系。在这种情况下，它使用相似方法之间的调用来定义两个不同类型的不同对象之间的关系。命名的相似性（因此假定的意图）为它们的关系性质提供了线索。
As with Delegation (and indeed all the Method Invocation EDPs), Retrieve describes a binary relationship between two objects and their enclosed methods. In this case, it defines a relationship between two dissimilar objects of dissimilar types, using a call between similar methods. The similarity of the naming, and therefore assumed intent, provides a clue to the nature of their relationship.
结果
尽管Retrieve与Delegation模式几乎完全相同，但方法相似的这个看似很小的额外要求却具有深远的影响。我们将在后面的模式中看到，当此模式与其他 EDP 和类型信息相结合时，可以快速形成复杂的交互并进行简单的描述。
Even though Retrieve is almost identical to the Delegation pattern, the seemingly small additional requirement that the methods be similar has some far-reaching effects. We will see in later patterns that when this pattern is combined with other EDPs and typing information, complex interactions can quickly be formed and described simply.
通过利用两个方法具有相同的名称以及这是声明方法意图的常用方法这一事实，我们可以推断出这两个方法具有共同的功能意图。此外，很明显，这是一种合适的方式，可以表明我们的原始调用站点方法正在请求被调用的方法执行部分工作，并且该工作与原始方法的核心功能紧密相关。
By leveraging the facts that both methods have the same name and that this is a common way of declaring the intent of a method, we can deduce that the two methods have a common functional intent. Furthermore, it becomes obvious that this is an appropriate way to indicate that our originating call-site method is requesting the invoked method to do some portion of work, and that work is tightly related to the core functionality of the original method.
被调用的对象可能驻留在调用方法中或由调用方法拥有，该对象包含调用方法，或者可能通过使用Retrieve来自其他地方。 实现部分显示了多种形式。
The object being called upon may reside in or be owned by the calling method, the object enclosing the calling method, or it may come from elsewhere via a use of Retrieve. The Implementation section shows multiple forms.
如果操作得当，此行为可用于形成继承的临时变体，方法是使用具有单个目标对象的多个重定向Redirectee实例将一个对象缝合到具有相同（或非常相似）接口的另一个对象。调用方法可能只是传递到，模仿继承的实现，或者它们可以提供自己的方法主体，模仿继承实现的覆盖，但会破坏重定向。如果它们都提供自己的方法主体，并在该主体中调用相应的目标方法Redirectee，那么这不仅是重定向，而且也是相关扩展方法的临时形式。
If done carefully, this behavior can be used to form an ad hoc variant of Inheritance by stitching one object to another with the same (or very similar) interface using multiple Redirection instances with a single target object. The calling methods may just pass through to the Redirectee, mimicking an inherited implementation, or they may provide their own method body, mimicking the overriding of an inherited implementation, but breaking the Redirection. If they both provide their own method body, and in that body call the appropriate target method in the Redirectee, then not only is this a Redirection, but it is also an ad hoc form of the related Extend Method.
执行
在 Java 中：
In Java:
public class Foo {
2 public void operation();
};
4
public class Bar {
6 Foo f;
public void operation() {
8 Foo f2;
f.operation(); // 重定向
10 f2.operation(); // 重定向
};
12 };
public class Foo {
2 public void operation();
};
4
public class Bar {
6 Foo f;
public void operation() {
8 Foo f2;
f.operation(); // Redirect
10 f2.operation(); // Redirect
};
12 };
在 Objective-C 中，说明如何使用Retrieve获取对 Redirectee 的引用：
In Objective-C, illustrating the use of Retrieve to get the reference to the Redirectee:
@interface Foo
2 {
}
4 -( void ) operation;
@end
6
@implementation Foo
8 -( void ) operation {
// 执行工作
10 };
@end
12
@interface Goo
14 {
Foo* f;
16 }
-(Foo*) getFoo;
18 @end
20 @implementation Goo
-(Foo*) getFoo {
22 return f;
};
24 @end
26 @interface Bar
{
28 Goo* g;
}
30 -( void ) operation;
@end
32
@implementation Bar
34 -( void ) operation {
[[g getFoo] operation]; // 检索时重定向
36 }
@end
@interface Foo
2 {
}
4 -(void) operation;
@end
6
@implementation Foo
8 -(void) operation {
// Do work
10 };
@end
12
@interface Goo
14 {
Foo* f;
16 }
-(Foo*) getFoo;
18 @end
20 @implementation Goo
-(Foo*) getFoo {
22 return f;
};
24 @end
26 @interface Bar
{
28 Goo* g;
}
30 -(void) operation;
@end
32
@implementation Bar
34 -(void) operation {
[[g getFoo] operation]; // Redirect on Retrieve
36 }
@end
相关模式
重定向是另一种非常通用的 EDP。它与委托的不同之处仅在于，方法现在具有意图和命名的相似性。改变这种相似性，您就会回到委托。另请参阅检索，了解如何通过加强对象类型相似性让目标对象在运行时发生变化。与委托一样，仅将对象相似性更改为相似，同时将对象类型相似性保留为不相似会导致未知状态。
Redirection is another very general EDP. It differs from Delegation only in that the methods now have a similarity of intent and naming. Alter that similarity, and you’re back to Delegation. Also see Retrieve for advice on how to let the target object vary at runtime by tightening the object type similarity. As with Delegation, changing just the object similarity to similar while leaving the object type similarity as dissimilar results in an unknown state.
方法调用分类
对象：不同
Object: Dissimilar
对象类型：不同
Object Type: Dissimilar
方法：类似
Method: Similar
集团
Conglomeration
对象行为
Object Behavioral
意图
将不同的操作和行为汇集或聚合在一起以便在单个对象内完成更复杂的任务。
To bring together, or conglomerate, diverse operations and behaviors in order to complete a more complex task within a single object.
也称为
分解消息，辅助方法
Decomposing Message, Helper Methods
动机
对象经常被要求执行一项任务，而这项任务太大或太难处理，无法在单个方法中完成。通常，将任务分解为多个小部分，以离散方法单独处理，然后由负责较大任务的方法将它们重新构建为一个整体结果，这在概念上是合理的。Kent Beck 将此称为分解消息模式 [ 5 ]。也可能将相关的子任务统一到单个方法中，从而实现单个对象内代码的重用。
An object is often asked to perform a task that is too large or unwieldy to be performed within a single method. Usually, it makes conceptual sense to break the task into smaller parts to be handled individually as discrete methods, and then build them back into a whole result by the method responsible for the larger task. Kent Beck refers to this as the Decomposing Message pattern [5]. It may also happen that related subtasks can be unified into single methods, resulting in the reuse of code inside a single object.
通过聚合重组较小的操作有几个好处。它通过让每个方法执行较少的工作来细化可触发的操作的粒度。这可以提高可维护性，因为允许在多个地方重复使用这些方法，而无需在多个地方复制和粘贴代码。此外，如果这些细粒度的方法公开可见，其他对象可能能够以新的方式使用部分行为。
The recomposition of smaller actions by Conglomeration has several benefits. It refines the granularity of actions that can be triggered by having each method perform less work. This can improve maintainability by allowing these methods to be reused in multiple places without copying and pasting the code in several places. Additionally, other objects may be able to use portions of the behavior in new ways if these fine-grained methods are made publicly visible.
考虑清单 5.17中的示例，它基于来自“重定向”的一个代码示例。我们认识到粉刷房间的方法实在是太长了；要采取的步骤太多，如果将其分解为较小的任务，代码将更容易阅读。我们将子任务方法公开，因为很明显，布置和清理房间的操作很有价值，可以独立于粉刷房间的操作进行请求。现在，他们CarefulPainter可以在天花板修补工或石膏板打磨工进来之前做好准备，也可以在之后进行清理。曾经是paintRoom行为中不可分割的一部分，现在被分解成有用的较小动作。
Consider the example in Listing 5.17, which builds on one of the code examples from Redirection. We recognized that the method for painting a room is simply too long; there are too many steps to be taken, and the code is easier to read if it is broken up into smaller tasks. We make the subtask methods publicly available because it becomes obvious that the actions of setting up and cleaning up a room are valuable and can be requested independently of the act of painting a room. Now, the CarefulPainter can do the preparation before a ceiling refinisher or a drywall sander comes in and can also perform the cleanup afterwards. What was once inextricably part of the paintRoom behavior is now broken out into useful smaller actions.
Listing 5.17. Prep work and cleanup are decomposable.
1 class SloppyFastPainter {
public:
3 void paintRoom( Room r ) {
// 粉刷墙壁
5 };
};
7
class CarefulPainter {
9 SloppyFastPainter sloppy;
Room currentTarpRoom;
11 Room currentHardwareRoom;
public:
13 void paintRoom( Room r ) {
this ->laydownTarps( r ); // 聚集
15 this ->removeHardware( r ); // 聚集
sloppy.paintRoom( r );
17 this ->replaceHardware(); // 聚集
this ->cleanUp(); // 聚集
19 };
void laydownTarps( Room r ) {
21 this ->currentTarpRoom = r;
// 铺开防水布
23 };
void cleanUp() {
25 // 清理 currentTarpRoom
// 放弃 currentTarpRoom
27 };
void removeHardware( Room r ) {
29 this ->currentHardwareRoom = r;
// 移除硬件
31 };
void replaceHardware() {
33 // 更换 currentHardwareRoom 中的硬件
// 放弃 currentHardwareRoom
35 };
};
1 class SloppyFastPainter {
public:
3 void paintRoom( Room r ) {
// Paint the walls
5 };
};
7
class CarefulPainter {
9 SloppyFastPainter sloppy;
Room currentTarpRoom;
11 Room currentHardwareRoom;
public:
13 void paintRoom( Room r ) {
this->laydownTarps( r ); // Conglomeration
15 this->removeHardware( r ); // Conglomeration
sloppy.paintRoom( r );
17 this->replaceHardware();// Conglomeration
this->cleanUp(); // Conglomeration
19 };
void laydownTarps( Room r ) {
21 this->currentTarpRoom = r;
// Laydown tarps
23 };
void cleanUp() {
25 // Clean currentTarpRoom
// Relinquish currentTarpRoom
27 };
void removeHardware( Room r ) {
29 this->currentHardwareRoom = r;
// Remove hardware
31 };
void replaceHardware() {
33 // Replace hardware in currentHardwareRoom
// Relinquish currentHardwareRoom
35 };
};
在这样做时，我们必须做出决定。我们可以允许将房间作为参数发送给每个子任务。这可以提供最大的灵活性，但让我们假设这里有雇主和工会的规定，其中一条规定是，出于问责制的原因，同一个人必须执行准备和完成任务。这可以防止在移除过程中丢失硬件被归咎于更换硬件的人，等等。每个人都完成他或她发起的工作。但是，不正确地跟踪这些数据可能会导致混乱。
In doing so, we had a decision to make. We could have allowed the room to be sent along as a parameter to each and every of the subtasks. This allows for the greatest flexibility, but let’s assume for a moment that there are employer and union regulations at work here, and one is that, for accountability reasons, the same person must perform the preparation and finalization tasks. This prevents, say, hardware lost during removal from being blamed on the person replacing it, and so on. Each person finishes the job he or she initiated. Not properly tracking this data can, however, lead to confusion.
如果一个没有组织的现场经理让几个CarefulPainter人跑来跑去，并负责告诉他们每个步骤要去哪个房间，他们可能会不小心派一个人去打扫另一个人已经准备好的房间。相反，我们选择让每个人知道房间信息CarefulPainter。想象一下，如果现场经理告诉工人去laydownTarps并递给他或她一张写有房间号的纸条。纸条放在工人的口袋里，当被要求时cleanUp，他或她只需参考纸条以确保打扫了正确的房间。完成后，工人交出纸条并准备去另一个房间。纸条代表的数据对工人来说是私密的，并将工人可以被要求执行的不同任务联系在一起。
If an unorganized site manager had several CarefulPainters running around and was responsible for telling them which room to go do each step in, they could accidentally send one to clean up a room that another had prepped. Instead, we choose to make the room information private to each CarefulPainter. Think of this as if the site manager tells a worker to go laydownTarps and hands him or her a chit with the room number. The chit is kept in the worker’s pocket, and when told to cleanUp, he or she simply refers to the chit to make sure the right room gets cleaned up. Once completed, the worker relinquishes the chit and is ready for another room. The data the chit represents is private to the worker and ties together separate tasks that worker can be asked to do.
聚合功能非常强大，但如果使用过度，则可能会引发问题。不合逻辑的结论是将每个语句都放在自己的方法中，但显然这不是一个好主意。任何时候应用聚合时，都必须在精细粒度的必要性和直接表达代码的简化和效率之间取得平衡。要找到合适的“原子行为”，就需要知道原子必须有多大。
Conglomeration is powerful, but it can cause issues if taken too far. The illogical conclusion would be to place every statement within its own method, but it should be obvious why that’s a bad idea. Any time Conglomeration is applied, a balance must be struck between the necessity for fine granularity and the simplification and efficiency of directly expressed code. Finding the proper “atomic behavior” requires knowing how large your atoms must be.
适用性
在以下情况下使用聚合：
Use Conglomeration when:
• 大任务可以分解为较小的子任务。
• A large task can be broken into smaller subtasks.
• 子任务必须在单个对象实例上执行，通常是由于共享的私有数据或状态。
• The subtasks must be performed on a single object instance, usually due to shared private data or state.
• 可以将多个子任务统一到一个方法体中。
• Several subtasks may be unified into a single method body.
结构
參與者
集团公司
Conglomerator
封闭对象类型。
Enclosing object type.
手术
operation
掌握分配子任务的控制方法。
Master controlling method that parcels out subtasks.
执行特定子任务的从属方法。
Subservient method performing a particular subtask.
合作
在Delegation的特殊化中，对象调用其自身的方法。调用点是operation，被调用的方法是operation2。
In a specialization of Delegation, the object calls a method of itself. The calling site is operation, and the method being called is operation2.
结果
与委托类似，此模式将两个方法绑定到依赖关系中，其中操作依赖于操作2的行为和实现。在这种情况下，与委托不同，它们共享的对象范围内的共享数据可能会立即产生副作用。
As with Delegation, this pattern ties two methods into a reliance relationship in which operation relies on the behavior and implementation of operation2. In this case, unlike Delegation, there may be immediate side effects on shared data within the confines of the object that they share.
执行
在 Java 中：
In Java:
public class Conglomerate {
2 public void operation() {
// 可选的准备工作
4 operation2();
// 可选的完成工作
6 };
public void operation2() {};
8 }
public class Conglomerate {
2 public void operation() {
// Optional prep work
4 operation2();
// Optional finish work
6 };
public void operation2() {};
8 }
在 Python 中：
In Python:
class Conglomerate:
2 def operation(self):
# 可选的准备工作
4 self.operation2();
# 可选的完成工作
6 def operation2(self):
# 请求的行为
8 pass
class Conglomerate:
2 def operation(self):
# Optional prep work
4 self.operation2();
# Optional finish work
6 def operation2(self):
# Requested behavior
8 pass
相关模式
Conglomeration将单个对象中的方法绑定在一起，因此是我们看到的第一个不具有依赖于Retrieve形式的 EDP 。修改对象相似性，以便调用另一个对象，同时保留相同的对象类型相似性，从而使用委托 Conglomeration。反转此操作以保留对象相似性，并更改对象类型与子类型关系的相似性会导致“还原方法”。这听起来可能有点奇怪——具有相同的对象但类型不同——但请阅读“还原方法”规范，了解为什么这不仅是可能的，而且非常有用。另一方面，保持对象相似性但将对象类型相似性更改为完全不同，在这种情况下是不合时宜的。最后，将方法相似性更改为对类似方法的调用会导致“递归”。
Conglomeration ties methods within a single object together and therefore is the first EDP we’ve seen that doesn’t have a form that relies on Retrieve. Modifying the object similarity so that the call is to another object, while retaining the same object type similarity, results in a use of Delegated Conglomeration. Reversing this to retain the object similarity, and changing the object type similarity to a subtyping relationship leads to Revert Method. This may sound a bit bizarre—having the same object but not the same type—but read the Revert Method specification for why this is not only possible but also highly useful. On the other hand, keeping the object similarity but changing the object type similarity to completely dissimilar is, at this juncture, ill formed. Finally, changing the method similarity such that the call is to a similar method results in Recursion.
当对象内部分解任务时，任何地方都会出现聚集现象，但有关此 EDP 的独特用法，请参见第 7 章中的模板方法。
Conglomeration will appear anywhere that an object is breaking down a task internally, but for a rather unique use of this EDP, see Template Method in Chapter 7.
方法调用分类
对象：类似
Object: Similar
对象类型：类似
Object Type: Similar
方法：不同
Method: Dissimilar
递归
Recursion
对象行为
Object Behavioral
意图
通过使用相同的对象状态执行许多较小且相似的任务来完成较大的任务。
To accomplish a larger task by performing many smaller and similar tasks while using the same object state.
动机
有时，经过深思熟虑，我们发现问题可以分解为与原始任务相同的子任务，只是规模较小。使用合并排序算法对项目数组进行排序就是这样一个例子。合并排序将一个数组分成两半，分别对每个数组进行排序，然后将两个排序后的数组合并为一个统一的整体。这两个半问题也使用合并排序进行排序，因此它们属于问题的相同一半，依此类推。最终，达到单个项目的数组，此时合并开始。
Sometimes, after consideration, we find that a problem can be broken down into subtasks that are identical to the original task, except on a smaller scale. Sorting an array of items using the merge sort algorithm is one such example. Merge sort takes an array and divides it into halves, sorting each individually, and then merges the two sorted arrays into a unified whole. The two half-problems are also sorted using merge sort, so they are subject to the same halving of the problem, and so on. Eventually, arrays of a single item are reached, at which point the merging begins.
方法调用自身的过程称为递归，它在一般编程中无处不在。同样的原则也适用于面向对象编程，但我们还有额外的要求，即对象必须通过隐式或显式使用self来调用自身。您几乎肯定熟悉这个概念，但它在 EDP 目录中有特定的位置和上下文。
The process by which a method calls itself is known as recursion, and it is ubiquitous in general programming. The same principle applies in object-oriented programming, but we have the additional requirement that the object must be calling on itself through implicit or explicit use of self. You’re almost certainly familiar with this concept, but it has a specific placement and context within the EDP catalog.
一般来说，递归是一种将大量计算折叠到小概念空间的方法。假设我们要对一个数组进行排序，我们决定对一个大数组进行排序的一种简单方法是将其拆分成两个大小相似的数组，然后在对两个子数组分别进行排序后合并它们。合并将很容易：如果数组 A 的头部小于数组 B 的头部，则将 A 的头部复制到新的更大的数组中；否则，复制 B 的头部。从相应的数组中删除复制的项目，并重复该过程，直到两个数组合并。
Generally speaking, recursion is a way of folding a large amount of computation into a small conceptual space. Assume we want to sort an array, and we decide that a simple way to sort a large array is to split it into two arrays of similar size and then merge the two subarrays after they are individually sorted. The merging will be easy: if the head of array A is less than the head of array B, then the head of A is copied to the new, larger array; otherwise, the head of B is copied. The copied item is removed from the appropriate array, and the process is repeated until both arrays are merged.
由于合并排序方案依赖于子数组的正确排序，因此我们正确地推测，我们可以通过依次拆分、排序然后合并两个子数组来对这两个子数组执行排序算法。我们再次面临相同的排序问题，因此我们以相同的方式继续，直到达到最小的不可分割数组：单个项目。此时，可以开始合并每个步骤中已排序的子数组的过程。
Because the merge sort scheme relies on the proper sorting of the sub-arrays, we correctly surmise that we can perform the sort algorithm on the two subarrays by splitting, sorting, and then merging each in turn. We are again faced with the same sorting problem, so we continue in the same manner until we reach the smallest indivisible array: a single item. At that point, the process of merging the sorted subarrays at each step can begin.
假设我们有一个包含和 的Array常用方法的类。如果起始数组的长度为 4 个项目，那么我们可以像下面的伪代码一样对整个排序进行硬编码，记住在面向对象系统中，我们默认应该有一个封闭的类或对象。addremove
Assume we have an Array class that contains the usual methods of add and remove. If the length of the beginning array was 4 items, then we could hardcode the entire sorting as in the following pseudocode, remembering that in an object-oriented system we should have an enclosing class or object by default.
class ArrayOfLength4Library {
2 数组 sort_merge(数组 a) {
元素 a11 = a[0];
4 元素 a12 = a[1];
元素 a21 = a[2];
6 元素 a22 = a[3];
数组 a1, a2, res;
8 // 对前半部分进行排序
if (a11 < a12) {
10 a1.add(a11);
a1.add(a12);
12 } else {
a1.add(a12);
14 a1.add(a11);
}
16 // 对后半部分进行排序
if (a21 < a22) {
18 a2.add(a21);
a2.add(a22);
20 } else {
a2.add(a22);
22 a2.add(a21);
}
24 // 合并
if (a1[0] < a2[0]) {
26 res.add(a1[0]);
a1.remove(0);
28 } else {
res.add(a2[0]);
30 a2.remove(0);
}
32 if (a1[0] < a2[0]) {
res.add(a1[0]);
34 a1.remove(0);
} else {
36 res.add(a2[0]);
a2.remove(0);
38 }
if (a1.length == 0) {
40 res.add(a2[0]);
res.add(a2[1]);
42 }
if (a2.length == 0) {
44 res.add(a1[0]);
res.add(a1[1]);
46 }
如果(a1.length == 1 && a2.length == 1) {
48 res.add(a1[0]);
res.add(a2[0]);
50 }
返回a;
52 };
};
class ArrayOfLength4Library {
2 Array sort_merge(Array a) {
Element a11 = a[0];
4 Element a12 = a[1];
Element a21 = a[2];
6 Element a22 = a[3];
Array a1, a2, res;
8 // Sort first half
if (a11 < a12) {
10 a1.add(a11);
a1.add(a12);
12 } else {
a1.add(a12);
14 a1.add(a11);
}
16 // Sort second half
if (a21 < a22) {
18 a2.add(a21);
a2.add(a22);
20 } else {
a2.add(a22);
22 a2.add(a21);
}
24 // Merge
if (a1[0] < a2[0]) {
26 res.add(a1[0]);
a1.remove(0);
28 } else {
res.add(a2[0]);
30 a2.remove(0);
}
32 if (a1[0] < a2[0]) {
res.add(a1[0]);
34 a1.remove(0);
} else {
36 res.add(a2[0]);
a2.remove(0);
38 }
if (a1.length == 0) {
40 res.add(a2[0]);
res.add(a2[1]);
42 }
if (a2.length == 0) {
44 res.add(a1[0]);
res.add(a1[1]);
46 }
if (a1.length == 1 && a2.length == 1) {
48 res.add(a1[0]);
res.add(a2[0]);
50 }
return a;
52 };
};
虽然效率很高，但这个解决方案有一个明显的缺点：它仅限于长度为 4 的数组。一个更灵活的版本是使用循环。这里我们说明一个for循环实现，为简单起见，假设数组的长度是 2 的幂：
While highly efficient, this solution has an obvious drawback: it is limited to only working for arrays of length 4. A much more flexible version is one using looping. Here we illustrate a for loop implementation, assuming for simplicity that the length of the array is a power of 2:
1 class ArrayOfPower2LengthLibrary {
Array sort_array(Array a) {
3 // 将 a 切成子
数组 subarray[0][0] = a;
5 for (i = 1 to log_2(a.length())) {
for (j = 0 to 2^i) {
7 prevarray = subarray[i-1][floor(j/2)]
subarray[i][j] = prevarray.slice(
9 ((j mod 2) *
(prevarray.length() / 2)),
11 ((j mod 2) + 1) *
(prevarray.length() / 2))
13 }
}
15 // 对子数组进行排序并合并
for (i = log_2(a.length()) to 1) {
17 for (j = 2^i to 0) {
subarray[i-1][j] =
19 max(subarray[i][j],
subarray[j][i]);
21 }
}
23 return subarray[0][0];
};
1 class ArrayOfPower2LengthLibrary {
Array sort_array(Array a) {
3 // Slice a into subarrays
subarray[0][0] = a;
5 for (i = 1 to log_2(a.length())) {
for (j = 0 to 2^i) {
7 prevarray = subarray[i-1][floor(j/2)]
subarray[i][j] = prevarray.slice(
9 ((j mod 2) *
(prevarray.length() / 2)),
11 ((j mod 2) + 1) *
(prevarray.length() / 2))
13 }
}
15 // Sort subarrays and merge
for (i = log_2(a.length()) to 1) {
17 for (j = 2^i to 0) {
subarray[i-1][j] =
19 max(subarray[i][j],
subarray[j][i]);
21 }
}
23 return subarray[0][0];
};
此解决方案成功地使我们的算法更加灵活，但代价是，由于开销机制，算法几乎不可读。请注意，我们向内循环到基本情况（数组长度为一个单位），在每一步存储状态，然后使用先前存储的状态向外循环。这个进入、存储、使用存储和展开的过程正是函数调用所完成的。我们可以利用这一事实，使用递归大大简化我们的实现：
This solution succeeds in making our algorithm much more flexible, but at a cost of making it almost unreadable because of the overhead mechanisms. Note that we are looping inward to the base case (length of array is a single unit), storing state at every step, then looping outward using the state previously stored. This process of enter, store, use store, and unroll is precisely what a function call accomplishes. We can use this fact to vastly simplify our implementation using Recursion:
class ArrayLibrary {
2 数组 sort (数组 a) {
return sort_array(a,
4 a.begIndex(), a.endIndex());
}
6 数组 sort_merge(数组 a, int beg, int end) {
if (a.length() > 1) {
8 数组 firstHalf =
sort_array(a, beg, end-beg/2);
10 // ^--- 递归 (隐式自身)
数组 secondHalf =
12 this ->sort_array(a,
(end-beg/2) + 1, end);
14 // ^--- 递归 (显式自身)
return merge(firstHalf, secondHalf);
16 } else {
return a;
18 }
};
20
Array merge(Array a, Array b) {
22 Array res;
while (b.length() > 1 || a.length() > 1) {
24 // 如果 a 或 b 为空，则 a[0] 或 b[0]
// 返回最小值
26 if (a[0] < b[0]) {
res.add(a[0]);
28 a. delete (0);
} else {
30 res.add(b[0]);
b. delete (0);
32 }
}
34 return res;
};
36 };
class ArrayLibrary {
2 Array sort (Array a) {
return sort_array(a,
4 a.begIndex(), a.endIndex());
}
6 Array sort_merge(Array a, int beg, int end) {
if (a.length() > 1) {
8 Array firstHalf =
sort_array(a, beg, end-beg/2);
10 // ^--- Recursion (implicit self)
Array secondHalf =
12 this->sort_array(a,
(end-beg/2) + 1, end);
14 // ^--- Recursion (explicit self)
return merge(firstHalf, secondHalf);
16 } else {
return a;
18 }
};
20
Array merge(Array a, Array b) {
22 Array res;
while (b.length() > 1 || a.length() > 1) {
24 // If a or b is empty, then a[0] or b[0]
// returns a min value
26 if (a[0] < b[0]) {
res.add(a[0]);
28 a.delete(0);
} else {
30 res.add(b[0]);
b.delete(0);
32 }
}
34 return res;
};
36 };
此版本在概念上更简洁，并且执行的任务与我们的循环变体相同。我们允许语言的运行时为我们处理开销。作为奖励，我们不再受某些长度的限制。代码不仅更简洁，而且高度通用。
This version is conceptually cleaner and performs the same task as our looping variation. We allow the runtime of the language to handle the overhead for us. As a bonus, we’re no longer limited to certain lengths. Not only is the code cleaner, it is highly generalized.
有时，例如在高性能嵌入式系统中，这种开销处理方式可能超出我们的需要。在这种情况下，将递归优化为循环或进一步优化为线性代码是可能的。然而，这些情况是极端的，而且往往是高度专业化的。在大多数情况下，递归的好处——概念清晰和代码简单——远远超过了速度的轻微损失。
There are times, such as in a high-performance embedded system, when that manner of overhead handling may be excessive for our needs. In such cases, optimizing away the recursion into loops, or further into linear code, is a possibility. These instances are extreme, however, and tend to be highly specialized. In most cases, the benefits of Recursion—conceptual cleanliness and simple code—greatly outweigh the small loss of speed.
适用性
在以下情况下使用递归：
Use Recursion when:
• 一个任务可以分成高度相似的子任务。
• A task can be divided into highly similar subtasks.
• 子任务必须或最好由同一个对象执行，通常是因为调用方法和被调用方法都需要访问公共的存储状态。
• The subtasks must be, or are preferred to be, performed by the same object, usually because of a need for access to common stored state by both the calling and called methods.
• 效率的微小损失被由此带来的简单性的提高所掩盖。
• A small loss of efficiency is overshadowed by the resulting gain in simplicity.
參與者
递归器
Recursor
Recursor具有在同一个实例中回调其自身的方法。
Recursor has a method that calls back on itself within the same instantiation.
operation是通过self.operation()调用自身的方法，无论是隐式还是显式。
operation is the method that calls itself, through self.operation(), whether implicitly or explicitly.
结构
合作
Recursor方法操作与自身协作，要求在每一步执行越来越小的任务，直到达到基准情况，此时结果被收集成最终结果。
Recursor’s method operation collaborates with itself, requesting that smaller and smaller tasks be performed at each step, until a base case is reached, at which point results are gathered into a final result.
结果
递归依赖正确形成的基线条件来终止递归堆栈是此模式的最薄弱环节。在大多数现代系统中，肆意消耗所有可用资源几乎是不可能的，但递归却可以通过具有永远不会满足的畸形基线条件轻松做到这一点。
The reliance of Recursion on a properly formed base case for termination of the recursion stack is the weakest point of this pattern. It is nearly impossible in most modern systems to wantonly consume all available resources, but recursion can do it easily by having a malformed base case that is never satisfied.
执行
在 Java 中：
In Java:
1 public class Recursor {
public void operation() {
3 // 可选的先前工作
self.operation();
5 // 可选的完成工作
};
7 }
1 public class Recursor {
public void operation() {
3 // Optional prior work
self.operation();
5 // Optional finish work
};
7 }
相关模式
递归可能是最不通用的方法调用 EDP，因为它指定对象、对象类型和方法相似性都设置为相似。每个都可以放宽以达到新的 EDP。聚合是通过使用同一对象和对象类型中的其他方法，同时使用与调用者不同的方法。（sort_merge前面的最后一个示例包含Conglomerationsort的实例，包括从到 的调用和从到 的sort_merge调用。）保留方法相似性但允许调用发生在同一对象类型的其他实例上会导致重定向递归，这是对象链中常见的概念。递归最有趣的邻居 EDP 可能是扩展方法，当方法和对象相似性得到保留但对象类型相似性设置为子类型关系时会遇到它。sort_mergemerge
Recursion is perhaps the least general method-call EDP, in that it specifies that object, object type, and method similarity are all set to similar. Each can be relaxed to reach new EDPs. Conglomeration is achieved by using other methods within the same object and object type while using methods that are dissimilar to the caller. (The final sort_merge example from earlier contains instances of Conglomeration, including the call from sort to sort_merge and the call from sort_merge to merge.) Retaining the method similarity but allowing the call to occur to other instances of the same object type results in a Redirected Recursion, a concept commonly seen in chains of objects. Perhaps the most interesting neighbor EDP of Recursion is Extend Method, which is encountered when the method and object similarity are preserved but the object type similarity is set to a subtyping relationship.
最后，委托还有一种特殊用途：两个对象通过两种不同的方法相互调用，共同完成一项任务，但私有数据是分开的。对象 A 的方法f()调用对象 B 的方法g()，进而g()调用对象 A 的方法。f()这称为相互递归，因为每个方法都间接调用自身。虽然这不是递归的真正示例，但从概念上讲，它足够相似，值得一提。
Finally, there is a specialized use of Delegation: when two objects mutually call each other via two different methods to work through a task together, but with separated private data. Object A’s method f() calls method g() on object B, and in turn g() calls method f() on object A. This is known as mutual recursion because each method calls itself indirectly. While not a true example of Recursion, it is conceptually similar enough to warrant mentioning.
方法调用分类
对象：类似
Object: Similar
对象类型：类似
Object Type: Similar
方法：类似
Method: Similar
恢复方法
Revert Method
对象行为
Object Behavioral
意图
绕过当前类的方法实现，而使用超类的实现，恢复为不太专业的方法体。
Bypass the current class’s implementation of a method, and use the superclass’s implementation instead, reverting to less specialized method body.
动机
多态性有时会对我们不利。在继承的模式规范中，您了解了减少重用实现期间的维护问题是该模式的驱动因素。应用于方法基本实现的修复会自动应用于子类，但子类可以覆盖现有方法或通过扩展方法添加新功能。然而，我们并不总是希望修复自动传播，而且我们并不总是调用类似的方法。有时我们需要使用继承树中更高层次的方法的先前实现，而不是类定义本地的实现。
Polymorphism sometimes works against us. In the pattern specification for Inheritance, you learned how reducing maintenance issues during reuse of an implementation is a driving factor for that pattern. Fixes applied to the base implementation for a method are auto applied to subclasses, but subclasses are allowed either to override existing methods or add new functionality through Extend Method. We don’t always want fixes to be propagated automatically, however, and we’re not always calling a similar method. Sometimes we need to use prior implementations of a method from higher up our inheritance tree instead of the one local to our class definition.
一个例子是提供多个版本的类供系统内同时使用。想象一下 Internet 数据传输协议的类库。基础库以 1.0 版本发布。使用 1.1 库，对底层协议进行了更改，并且使用 1.1 协议的应用程序在检测到连接另一端的应用程序仅启用 1.0 版本时必须能够回退到 1.0 协议。
One such instance is when providing multiple versions of classes for simultaneous use within a system. Imagine a library of classes for an Internet data transfer protocol. A base library is shipped as 1.0. With the 1.1 library, changes to the underlying protocol are made, and an application using the 1.1 protocol must be able to fall back to the 1.0 protocol when it detects that the application at the other end of the connection is only 1.0 enabled.
一种方法是直接使用多态性，并定义一个抽象提供协议方法的基类（如图 5.1所示），并在协议检测时创建适当的类项。不幸的是，这种方法没有提供很多动态灵活性，并且不允许协议动态适应不断变化的需求。
One approach would be to use polymorphism directly and define a base class that abstractly provides the protocol’s methods, as in Figure 5.1, and create the proper class item on protocol detection. Unfortunately, this approach does not offer a lot of dynamic flexibility and would not allow a protocol to adapt on the fly to changing demands.
图 5.1. 多态方法
Figure 5.1. Polymorphic approach
虽然它使使用旧协议与客户端建立连接变得简单，但在发生错误时，它使连接的正常降级变得更加困难。假设您正在传输视频流，如果网络条件良好且延迟较低，则可以采用新的处理器密集型数据压缩算法。如果但是，如果延迟较高，则另一端解压缩所需的 CPU 时间会导致跳过。如果您发现网络拥塞，则可以使用较旧的压缩系统。
Although it makes setting up a connection with client using the older protocol simple, it makes graceful degradation of the connection in case of an error more problematic. Let’s say you’re transmitting a video stream, and if network conditions are good and latency is low, then a new processor-intensive compression algorithm for the data can be brought to bear. If latency is high, however, then the CPU time needed for decompression at the other end would cause skipping. If you notice that the network is congested, you drop down to the older compression system.
您可以实例化每种协议类型的对象，并根据需要切换到早期版本，但可能会出现协议状态问题，并带来麻烦。从一个协议处理程序类转换为另一个协议处理程序类需要实例化新类的新对象，然后在每次可能需要进行协议转换时复制所有私有数据。希望这些数据可以专门保存在ProtocolBase类中，但即使这样，也必须打破私有数据的严格封装才能完成复制。此外，这会将协议知识分割在Protocol类和类之间，如清单 5.18Controller所示。前者类实现实际行为，但后者必须知道每种行为何时合适。
You could instantiate objects of each protocol type and swap to earlier versions as needed, but protocol state issues might arise and be troublesome. Converting from one protocol handler class to another would require instantiating a new object of the new class and then copying over any private data each and every time a possible protocol shift was needed. Hopefully this data can be held exclusively in the ProtocolBase class, but even then there has to be a breaking of strict encapsulation of the private data to accomplish the copy. Further, this splits the protocol knowledge between the Protocol classes and the Controller class, as shown in Listing 5.18. The former classes implement that actual behavior, but the latter has to know when each behavior is appropriate.
Listing 5.18. Instance swapping for protocol fallback in C++.
1 class CommonData{};
class RawData{};
3 class Video{};
5 class Connection{
public:
7 bool lowLatency(); // 网络状态良好吗？
void transmission(RawData);
9 };
11 class RemoteEnd {
public:
13 int protocolVersion();
};
15
// 协议处理程序类：
17 class ProtocolBase {
protected:
19 CommonData data;
int d_version;
21 Connection connection;
public:
23 virtual
Connection initConnection(RemoteEnd otherSide);
25 Connection getConnection() {
return connection;
27 };
virtual void sendData(Video vid) {
29 connection.transmit(compress(vid));
};
31 int version() { return d_version; };
CommonData getData() { return data; };
33 virtual RawData compress(Video vid) = 0;
};
35
class Protocol1dot0 : public ProtocolBase {
37 public:
Protocol1dot0() {
39 d_version = 1.0;
};
41 Protocol1dot0(ProtocolBase* base) {
d_version = 1.0;
43 this ->data = base->getData();
};
45 RawData compress(Video vid);
};
47
类Protocol1dot1 :公共ProtocolBase {
49 公共：
Protocol1dot1() {
51 d_version = 1.1;
};
53 Protocol1dot1(ProtocolBase* base) {
d_version = 1.1;
55 this ->data = base->getData();
}
57 RawData 压缩(Video vid);
};
59
类控制器 {
61 ProtocolBase* handler;
公共：
63 void setupConnection(RemoteEnd otherSide) {
if (otherSide.protocolVersion() == 1.0) {
65 handler = new Protocol1dot0();
} else {
67 handler = new Protocol1dot1();
}
69 handler->initConnection(otherSide);
};
71 void sendData(Video vid) {
bool networkGood =
73 handler->getConnection().lowLatency();
if (networkGood &&
75 handler->version() == 1.0) {
// 向前倒下
77 handler = new Protocol1dot1(handler);
} else if (!networkGood &&
79 handler->version() == 1.1) {
// 回退
81 handler = new Protocol1dot0(handler);
}
83 处理程序->sendData(vid);
};
85 };
1 class CommonData{};
class RawData{};
3 class Video{};
5 class Connection{
public:
7 bool lowLatency(); // is network in good shape?
void transmit(RawData);
9 };
11 class RemoteEnd {
public:
13 int protocolVersion();
};
15
// Protocol handler classes:
17 class ProtocolBase {
protected:
19 CommonData data;
int d_version;
21 Connection connection;
public:
23 virtual
Connection initConnection(RemoteEnd otherSide);
25 Connection getConnection() {
return connection;
27 };
virtual void sendData(Video vid) {
29 connection.transmit(compress(vid));
};
31 int version() { return d_version; };
CommonData getData() { return data; };
33 virtual RawData compress(Video vid) = 0;
};
35
class Protocol1dot0 : public ProtocolBase {
37 public:
Protocol1dot0() {
39 d_version = 1.0;
};
41 Protocol1dot0(ProtocolBase* base) {
d_version = 1.0;
43 this->data = base->getData();
};
45 RawData compress(Video vid);
};
47
class Protocol1dot1 : public ProtocolBase {
49 public:
Protocol1dot1() {
51 d_version = 1.1;
};
53 Protocol1dot1(ProtocolBase* base) {
d_version = 1.1;
55 this->data = base->getData();
}
57 RawData compress(Video vid);
};
59
class Controller {
61 ProtocolBase* handler;
public:
63 void setupConnection(RemoteEnd otherSide) {
if (otherSide.protocolVersion() == 1.0) {
65 handler = new Protocol1dot0();
} else {
67 handler = new Protocol1dot1();
}
69 handler->initConnection(otherSide);
};
71 void sendData(Video vid) {
bool networkGood =
73 handler->getConnection().lowLatency();
if (networkGood &&
75 handler->version() == 1.0) {
// fall forward
77 handler = new Protocol1dot1(handler);
} else if (!networkGood &&
79 handler->version() == 1.1) {
// fall back
81 handler = new Protocol1dot0(handler);
}
83 handler->sendData(vid);
};
85 };
在每个方法中都放置大量的开销Controller。我们可以尝试将所有这些控制器逻辑放入类中ProtocolBase，乍一看似乎可行。但是，它要求类ProtocolBase了解所有子类，这通常是一种相当糟糕的设计。这也意味着ProtocolBase必须在定义新协议时更新。如果此类包含在预编译库中供客户端使用，则客户端无法扩展该库以适应新协议版本。
This is a tremendous amount of overhead being placed in each of the methods in Controller. We could have tried to put all of this controller logic in to the ProtocolBase class, and at first glance that looks like it could work. However, it requires that the ProtocolBase class know about all of the subclasses, which is fairly poor design in general. It also means that ProtocolBase must be updated anytime a new protocol is defined. If this class is shipped in a precompiled library for clients to use, then the clients cannot extend the library for new protocol versions.
此外，这种方法在实际实现中不易实现代码重用。如果协议 1.1 只是对协议 1.0 的一个小调整，那么如果我们可以重用 1.0 的行为就好了，这可能通过将通用代码提升到基类并允许从每个子类调用它来实现。这种方法可能适用于两个子类。然而，一旦我们开始涉及更多的子类、协议版本和特殊情况，它可能会变得繁琐。可定义通用代码核心的可能性会迅速下降。
In addition, this approach doesn’t easily allow for code reuse for the actual implementation. If Protocol 1.1 is just a small tweak off of Protocol 1.0, then it would be nice if we could reuse 1.0’s behavior, which would probably be done by hoisting the common code into the base class and allowing it to be called from each subclass. This approach is likely to work well for two subclasses. Once we start getting into more subclasses, protocol versions, and special cases, however, it is likely to become cumbersome. The likelihood of a common core of code being definable goes down quickly.
很快就会发现，这种方法不能很好地支持优雅的动态回退和前进。
It fast becomes apparent that graceful dynamic fallback and fallfoward is not well supported by this approach.
相反，我们可以让每个新协议版本都从先前的协议版本中子类化，这样只需要对协议进行必要的更改。这样，当我们首次确定每个端点可能支持的最佳协议时，我们只需要处理一个类，但我们仍然需要能够恢复到以前的版本。图 5.2显示了这种方法的扩展变体，包括多个版本。
Instead, we can have each new protocol version subclass from the prior protocol version so that only the necessary changes to the protocol need to be implemented. This lets us have only a single class to deal with when we first figure out the best possible protocol that each end might be able to support, but we still need to be able to revert to the previous version. Figure 5.2 shows an extended variant of this approach, including several versions.
图 5.2. 子类化方法
Figure 5.2. Subclassing approach
清单 5.19中展示了这种方法的一个示例，其中只有两个类。支持类与清单 5.18中一样。在这种情况下，我们可以实例化链中最后一个类的对象，代码中的条件语句将在需要时将协议处理传递回链中的相应版本。这大大简化了支持此协议的代码的进一步维护，因为每当库发布新更新时，只需要将协议处理对象的实例更改为最新版本，因此所有优雅降级都会自动处理。
An example of this approach is shown in Listing 5.19 with just two classes. The supporting classes are as in Listing 5.18. In this case, we can instantiate an object of just the last class in the chain, and conditional statements in the code will pass the protocol handling back up the chain to the appropriate version if needed. This vastly simplifies further maintenance of the code that supports this protocol because only the instance(s) of the protocol-handling object need to be changed to the latest version anytime a new update to the library comes out, and all the graceful degradation is therefore handled automatically.
Listing 5.19. Auto fallback/forward using Revert Method.
1 class Protocol1dot0 {
public:
3 virtual
void initConnection(RemoteEnd otherSide) {
5 connection = otherSide.connect(1.0);
};
7 virtual
void sendData(Video vid) {
9 RawData rawData = compress(vid);
connection.transmit(rawData);
11 };
protected:
13 virtual
RawData compress(Video);
15 Connection connection;
CommonData data;
17 };
19 class Protocol1dot1 : public Protocol1dot0 {
public:
21 virtual
void initConnection(RemoteEnd otherSide) {
23 connection = otherSide.connect(1.1);
};
25 virtual
void sendData(Video vid) {
27 RawData rawData;
if (connection.lowLatency()) {
29 rawData = compress(vid);
} else {
31 rawData = Protocol1dot0::compress(vid);
// ^--- 恢复方法
33 }
connection.transmit(rawData);
35 };
protected:
37 virtual
RawData compress(Video);
39 };
1 class Protocol1dot0 {
public:
3 virtual
void initConnection(RemoteEnd otherSide) {
5 connection = otherSide.connect(1.0);
};
7 virtual
void sendData(Video vid) {
9 RawData rawData = compress(vid);
connection.transmit(rawData);
11 };
protected:
13 virtual
RawData compress(Video);
15 Connection connection;
CommonData data;
17 };
19 class Protocol1dot1 : public Protocol1dot0 {
public:
21 virtual
void initConnection(RemoteEnd otherSide) {
23 connection = otherSide.connect(1.1);
};
25 virtual
void sendData(Video vid) {
27 RawData rawData;
if (connection.lowLatency()) {
29 rawData = compress(vid);
} else {
31 rawData = Protocol1dot0::compress(vid);
// ^--- Revert Method
33 }
connection.transmit(rawData);
35 };
protected:
37 virtual
RawData compress(Video);
39 };
现在我们不再需要Controller和ProtocolBase类。任何客户端只需实例化它能够使用的最新版本，任何回退或前进都会自动发生。代码更加简洁，更直接地表达了开发人员的意图。
Now we have eliminated the need for the Controller and ProtocolBase classes. Any client only needs to instantiate the newest version that it is capable of working with, and any fallback or fallforward happens automatically. The code is considerably cleaner and more directly expresses the intent of the developer.
适用性
在以下情况下使用恢复方法：
Use Revert Method when:
• 一个类希望重新启用超类对其已重写的方法的实现，以便使用原始行为。
• A class wishes to reenable a superclass’s implementation of a method that it has overridden in order to use the original behavior.
结构
參與者
原始行为
OriginalBehavior
至少定义方法operation2的基类。
A base class defining at least the method operation2.
手术
operation
操作是发生调用的方法，至少在RevertedBehavior类中定义。
operation is the method in which the call takes place and is defined in at least the RevertedBehavior class.
操作2
operation2
operation2是被调用的方法，并且至少在OriginalBehavior类中定义。如果 也必须在 的子类中定义OriginalBehavior，则在 中RevertedBehavior或在类层次结构中它们之间定义的类中。第二个定义会屏蔽 中的实现，使其OriginalBehavior不成为 中的默认实现RevertedBehavior，并强制选择恢复的实现。
operation2 is the method being called and is defined in at least the OriginalBehavior class. If must also be defined in a subclass of OriginalBehavior, either in RevertedBehavior or in a class defined between them in the class hierarchy. This second definition masks the implementation in OriginalBehavior from being the default within RevertedBehavior and forcing the selection of the reverted implementation.
的子类OriginalBehavior，定义了操作并且覆盖了操作2。当我们通常期望操作调用该方法自己的实现时，它却调用操作2Original-Behavior的实现。
A subclass of OriginalBehavior, with operation defined and operation2 overridden. operation calls the Original-Behavior implementation of operation2 when we would normally expect it to call its own implementation of that method.
合作
在大多数情况下，子类会重写父类的方法来替换其功能，但两个方法定义可以协同工作以允许扩展行为。RevertedBehavior依赖OriginalBehavior于核心实现。
In most cases, a subclass overrides a parent class’s method to replace its functionality, but the two method definitions can work together to allow an extension of the behavior. RevertedBehavior relies on OriginalBehavior for a core implementation.
结果
覆盖基类方法与使用相同方法之间存在概念上的脱节，这可能会让一些学生和从业者感到困惑。覆盖方法不会删除旧方法；它只是将旧方法隐藏在子类对象的公共视图之外。对象仍然知道其父类的方法，并且可以在内部调用它们，而不会将这些知识暴露给外部世界。
There is a conceptual disconnect between the overriding of a base class’s method and the utilization of that same method that can be confusing to some students and practitioners. Overriding a method does not erase the old method; it merely hides the old method from public view for objects of the subclass. The object still has knowledge of its parent’s methods and can invoke them internally without exposing this knowledge to the external world.
恢复方法是聚合的一种形式，其中添加了超类实现的知识和直接访问它的能力。
Revert Method is a form of Conglomeration to which the knowledge of a superclass’s implementation and the ability to access it directly have been added.
执行
在 C# 中：
In C#:
1 public class OriginalBehavior {
// 可选的 operation() 定义
3 public virtual void operation() {};
public virtual void operation2() {};
5 };
7 public class RevertedBehavior : OriginalBehavior {
public override void operation() {
9 // 可选的工作优先
if (oldBehaviorNeeded) {
11 OriginalBehavior::operation2();
} else {
13 operation2();
}
15 // 可选的工作之后
};
17 // 如果不覆盖此方法，则会出现
// 不需要恢复到超类的版本
19 public override void operation2();
};
1 public class OriginalBehavior {
// Optional definition of operation()
3 public virtual void operation() {};
public virtual void operation2() {};
5 };
7 public class RevertedBehavior : OriginalBehavior {
public override void operation() {
9 // Optional work priot
if (oldBehaviorNeeded) {
11 OriginalBehavior::operation2();
} else {
13 operation2();
}
15 // Optional work after
};
17 // If this were not overridden, there would be
// no need to revert to superclass's version
19 public override void operation2();
};
相关模式
如前所述，Revert Method与Conglomeration密切相关，不同之处在于它使用要聚合的方法的超类型实现。如果要调用类似的方法，则有一个Extend Method的示例。如果行为由 定义的超类型的其他对象提供OriginalBehavior，那么这就是Trusted Delegation的一个实例。
As noted earlier, Revert Method is closely related to Conglomeration except that it uses a supertype’s implementation of the method to be conglomerated. If the similar method is to be called, then you have an example of Extend Method. If the behavior is being provided by the other objects of the supertype defined by OriginalBehavior, then that is an instance of Trusted Delegation.
方法调用分类
对象：类似
Object: Similar
对象类型：子类型
Object Type: Subtype
方法：不同
Method: Dissimilar
扩展方法
Extend Method
对象行为
Object Behavioral
意图
在重用现有代码的同时，补充而不是替换超类方法中的行为。
Supplement, not replace, behavior in a method of a superclass while reusing existing code.
也称为
延长超级
Extending Super
动机
软件生产和维护中最常见的两项任务是修复错误或添加新功能。最直接的方法是直接修改代码。如果存在错误，就修复它。如果需要新功能，就添加它。
Two of the most common tasks in software production and maintenance are to fix bugs or add new features. The most straightforward way of doing so is to alter the code directly. Where the bug exists, fix it. Where the new feature is needed, add it.
然而，这并不总是可行的。通常，需要修复或增强的代码无法以源代码形式提供，因为它已作为库提供。其他时候，必须保留原始代码和行为以供遗留代码使用。无论哪种情况，有时原始方法都无法更改。
This isn’t always possible, however. Often the code that needs fixing or enhancing is not available in source form because it has been provided as a library. Other times, the original code and behavior must be preserved for use by legacy code. In either case, there will be times when the original method cannot be changed.
当然，如果原始源代码可用，我们可以将旧代码复制并粘贴到新方法中。但是，如继承规范中所示，这会带来许多问题，包括方法不一致，并导致潜在的维护困境。一种更不容易出错的方法是仅在一个点定义任何实现，然后重用现有代码，根据需要调整输入或结果。数据可以在传递给原始实现之前进行修改，或者可以使用原始实现的结果来执行其他步骤。遵守这一原则是减少维护复杂性和生成清晰代码的简单方法。
If the original source is available, of course, we can just copy and paste the old code into the new method. As shown in the specification for Inheritance, however, this presents a host of problems, including inconsistency of methods, and results in a potential maintenance morass. A much less error-prone approach is to have any implementation defined at only a single point and then reuse the existing code, tweaking the input or results as needed. Data can be massaged before being passed to the original implementation, or the results of the original implementation can be used to perform additional steps. Adhering to this principle is a simple way to reduce maintenance complexity and produce clear code.
如果原始源代码不可用，那么无论如何我们都无法复制和粘贴代码，而必须找到另一种方法来重用现有代码。当然，可以创建对具有原始行为的委托对象的引用，并在需要时调用它。这是Redirection中采用的方法，但这种方法在某些情况下很快就会失效，如清单 5.20所示。如果添加的行为需要可以访问原始代码封装为私有的数据，并且代码不会将其暴露给外部客户端使用或操作，那么这种策略根本行不通。即使可以，新代码的作用以及它与原始代码的关系可能并不明确。
If the original source is not available, then we’re prevented from copying and pasting code in any case and have to find another way to reuse the existing code. It is possible, of course, to create a reference to a delegate object with the original behavior and call into it when needed. This is the approach taken in Redirection, but this approach breaks down quickly in certain cases, as shown in Listing 5.20. If the added behavior needs to have access to data that was encapsulated as private by the original code, and the code does not expose it for use or manipulation by outside clients, then this tactic simply isn’t possible. Even when it is possible, it may not be clear what the new code is doing and that its relationship to the original code is quite close.
Listing 5.20. Using Redirection in Python to add behavior.
class OriginalBehavior:
2 __privateData = True
# 在私有属性前面添加 __
4 def desireBehavior(self, data):
# 做一些有趣的事情
6 return data
8 class NewBehavior:
def addMoreBehavior(self, data):
10 # 执行任何必要的调用前设置
# 无法访问 ob 的 __privateData
12 ob = OriginalBehavior()
val = ob.desiredBehavior(data)
14 # 调用后清理或更多功能
class OriginalBehavior:
2 __privateData = True
# Prepend __ for a private attribute
4 def desiredBehavior(self, data):
# Do something interesting
6 return data
8 class NewBehavior:
def addMoreBehavior(self, data):
10 # Perform any necessary pre-call setup
# Cannot access __privateData of ob
12 ob = OriginalBehavior()
val = ob.desiredBehavior(data)
14 # Post-call cleanup or more functionality
这些问题的解决方案是使用Extend Method，如清单 5.21 所示。我们使用继承来实现一个新类，该类从包含原始所需行为的类中派生出子类，因为它提供了重用原始代码的机制。然后，我们重写原始方法，添加我们需要的方法实现，但在需要执行其行为时回调超类的方法实现。这为我们提供了对更改后的行为的简单维护、重用和封装。
The solution to these problems is to use Extend Method, as in Listing 5.21. We implement a new class that subclasses off of the class containing the original desired behavior using Inheritance because it provides the mechanisms for reuse of the original code. We then override the original method, adding what implementation of the method we need, but make a call back to the superclass’s implementation of the method when we need its behavior to be executed. This provides us with simple maintenance, reuse, and encapsulation of the altered behavior.
Listing 5.21. Using Extend Method to add behavior.
class OriginalBehavior:
2 __privateData = True
# 在私有属性前面添加 __
4 def desireBehavior(self, data):
# 做一些有趣的事情
6 返回数据
8 class NewBehavior(OriginalBehavior):
def desireBehavior(self, data):
10 # 执行任何必要的调用前设置
# 可以访问 __privateData！
12 superDelegate = super(NewBehavior, self)
superDelegate.desiredBehavior(data)
14 # ^--- 扩展方法
# 调用后清理或更多功能
class OriginalBehavior:
2 __privateData = True
# Prepend __ for a private attribute
4 def desiredBehavior(self, data):
# Do something interesting
6 return data
8 class NewBehavior(OriginalBehavior):
def desiredBehavior(self, data):
10 # Perform any necessary pre-call setup
# Can access __privateData!
12 superDelegate = super(NewBehavior, self)
superDelegate.desiredBehavior(data)
14 # ^--- Extend Method
# Post-call cleanup or more functionality
适用性
在以下情况下使用扩展方法：
Use Extend Method when:
• 方法的现有行为需要扩展但不能被替换。
• Existing behavior of a method needs to be extended but not replaced.
• 由于缺少源代码，因此需要重复使用代码。
• Reuse of code is preferred or necessitated by lack of source code.
• 使用重定向是不可能的或者不是最佳的，因为新的行为需要访问原始实现的私有数据。
• Using Redirection is not possible or optimal because the new behavior needs access to data that is private to the original implementation.
结构
參與者
原始行为
Original Behavior
定义接口并包含具有所需核心功能的方法操作的实现。
Defines interface and contains an implementation of the method operation with the desired core functionality.
手术
operation
由 定义以OriginalBehavior提供基本行为。由 重写以ExtendedBehavior提供利用原始实现的新实现。
Defined by OriginalBehavior to provide basic behavior. Overridden in ExtendedBehavior to provide a new implementation that makes use of the original implementation.
扩展行为
Extended Behavior
使用接口OriginalBehavior，然后重新实现操作方法，包括对基类代码的调用，并围绕其添加的代码和/或行为。
Uses interface of OriginalBehavior, then reimplements the operation method to include a call to the base class code surrounded by added code and/or behavior.
合作
在大多数情况下，当子类重写父类的方法时，目的是替换原始方法的功能。但是，这两个方法定义可以协同工作，以允许扩展父类的行为，而不是替换。ExtendedBehavior依赖OriginalBehavior于接口和核心实现。
In most cases when a subclass overrides a parent class’s method, the purpose is to replace the functionality of that original method. The two method definitions, however, can work together to allow an extension of the behavior of the parent class instead of a replacement. ExtendedBehavior relies on OriginalBehavior for both interface and core implementation.
结果
与Revert Method一样，调用方法的重写版本的概念可能会令人困惑。在Extend Method中，这可能会更加令人困惑，因为调用方法似乎在调用幽灵方法。使用此模式可以优化代码重用，但operation中的方法OriginalBehavior变得有些脆弱——现在 依赖于其行为ExtendedBehavior::Operation随时间不变。行为以多态方式扩展，并且对 的客户端透明OriginalBehavior。
As with Revert Method, the concept of calling an overridden version of a method can be confusing. In Extend Method, it can be even more confusing because the calling method seems to be invoking a ghost method. Using this pattern optimizes code reuse, but the method operation in OriginalBehavior becomes somewhat fragile—its behavior is now relied on by ExtendedBehavior::Operation to be invariant over time. Behavior is extended polymorphically and transparently to clients of OriginalBehavior.
执行
在 Java 中：
In Java:
1 public class OriginalBehavior {
public void operation() {
3 // 执行核心行为
};
5 };
public class ExtendedBehavior
7 extends OriginalBehavior {
public void operation() {
9 //
super .operation() 之前的可选工作;
11 // 之后的可选工作
};
13 };
1 public class OriginalBehavior {
public void operation() {
3 // Do core behavior
};
5 };
public class ExtendedBehavior
7 extends OriginalBehavior {
public void operation() {
9 // Optional work before
super.operation();
11 // Optional work after
};
13 };
在 C++ 中：
In C++:
1 class OriginalBehavior {
public:
3 virtual void operation() {
// 执行核心行为
5 };
};
7 class ExtendedBehavior : public OriginalBehavior {
public:
9 void operation() {
// 执行前的可选工作
11 OriginalBehavior::operation(); // 13
之后的可选工作 }; };
1 class OriginalBehavior {
public:
3 virtual void operation() {
// Do core behavior
5 };
};
7 class ExtendedBehavior : public OriginalBehavior {
public:
9 void operation() {
// Optional work before
11 OriginalBehavior::operation();
// Optional work after
13 };
};
相关模式
Extend Method与Recursion类似，使用单个对象进行定义。将此对象相似性转换为截然不同的对象将产生Trusted Redirection的实例，该实例使用多态性遍历一组受信任的相关类以实现功能。放宽方法相似性以调用超类的不同方法将产生Revert Method。由于Revert Method是Conglomeration的超类感知变体，因此Extend Method是Recursion的超类感知版本，尽管结果行为完全不同。只需删除该超类访问并调用当前类型的实现即可实现递归。
Extend Method, like Recursion, is defined using a single object. Converting this object similarity to distinct dissimilar objects results in an instance of Trusted Redirection, which uses polymorphism to traverse a cluster of trusted related classes for functionality. Relaxing the method similarity to call a different method of the superclass results in Revert Method. Because Revert Method is a superclass-aware variation of Conglomeration, Extend Method is a superclass-aware version of Recursion, even though the resulting behavior is quite different. Recursion can be reached by simply removing that superclass access and invoking the current type’s implementation.
方法调用分类
对象：类似
Object: Similar
对象类型：超级
Object Type: Super
方法：类似
Method: Similar
委托集团
Delegated Conglomeration
对象行为
Object Behavioral
意图
当多个同一类型的对象必须协同工作才能完成一项任务时，每个对象都执行不同的行为。
When multiple objects of the same type must work in concert to complete a task, with each performing different behaviors.
动机
相同类型的对象通常协同工作以执行任务。具有异质行为的同质数据环境通常被编码为一组协作以产生更大功能的类似对象。
Objects of the same type often work in concert to perform tasks. Homogeneous data environments with heterogeneous behavior are frequently coded as a collection of like objects collaborating to produce a larger functionality.
清单 5.22展示了一个使用社交网站的示例。每个用户的帐户都有一个好友列表，用户可能想要做的一件事就是邀请其他好友参加活动。用户可以设置自己的通知偏好，包括他们收到邀请时想要联系的电子邮件地址。
Listing 5.22 presents an example using a social networking site. Each user’s account has a list of friends, and one action users are likely to want to do is invite other friends to an event. Users can set their own notification preferences, including the email address they want to be contacted at when they get an invitation.
Listing 5.22. Inviting friends naively in Java.
导入java.util.Vector;
2 导入java.lang.String;
4 public class UserAccount {
Vector<UserAccount> allMyFriends;
6 public String notificationEmail;
8 public void invitationToEvent (
Vector<UserAccount> friends, String msg) {
10 for (UserAccount friend : friends) {
String email = friend.notificationEmail;
12 // 发送电子邮件
}
14 }
};
import java.util.Vector;
2 import java.lang.String;
4 public class UserAccount {
Vector<UserAccount> allMyFriends;
6 public String notificationEmail;
8 public void inviteToEvent (
Vector<UserAccount> friends, String msg) {
10 for (UserAccount friend : friends) {
String email = friend.notificationEmail;
12 // send email
}
14 }
};
此解决方案有效，但它要求每个帐户公开其电子邮件信息。大多数用户不想在社交网站上公开他们的电子邮件地址。为了让这些信息稍微保密一些，活动策划者可以向朋友索要电子邮件，朋友可以根据需要提供这些电子邮件，如清单 5.23所示。
This solution works, but it requires each account to expose its email information publicly. Most users do not want to expose their email address on a networking site. To keep this information slightly more private, event planners can request emails from friends, and friends can provide those emails if they wish to, as shown in Listing 5.23.
Listing 5.23. A slightly better approach for inviting friends.
1 导入java.util.Vector;
导入java.lang.String;
3
public class UserAccount {
5 Vector<UserAccount> allMyFriends;
public String notificationEmail;
7
public void invitationToEvent (
9 Vector<UserAccount> friends, String msg) {
for (UserAccount friend : friends) {
11 String email = friend.getEmail( this );
// 发送电子邮件
13 }
}
15
public String getEmail (UserAccount requester) {
17 String returnVal = "";
if (allMyFriends.contains(requester)) {
19 returnVal = notificationEmail;
}
21 return returnVal;
};
23 };
1 import java.util.Vector;
import java.lang.String;
3
public class UserAccount {
5 Vector<UserAccount> allMyFriends;
public String notificationEmail;
7
public void inviteToEvent (
9 Vector<UserAccount> friends, String msg) {
for (UserAccount friend : friends) {
11 String email = friend.getEmail(this);
// send email
13 }
}
15
public String getEmail (UserAccount requester) {
17 String returnVal = "";
if (allMyFriends.contains(requester)) {
19 returnVal = notificationEmail;
}
21 return returnVal;
};
23 };
这种方法仍然要求用户向好友列表中任何请求的人提供他们的电子邮件地址。同样，大多数用户可能只想收到邀请而不必分享他们的电子邮件地址。通过使用委托聚合，可以将电子邮件地址的可见性与接收邀请的能力分开，如清单 5.24 所示。现在，当用户邀请朋友时，他们不会直接看到或访问朋友的电子邮件信息。相反，他们会向朋友的帐户发送通知请求，这确保了邀请在不暴露电子邮件地址的情况下送达。
This approach still requires users to provide their email address to anyone on their friends list who requests it. Again, most users would probably just like to receive an invitation without having to share their email address. Visibility of an email address can be separated from the capability to receive invitations by use of Delegated Conglomeration, as in Listing 5.24. Now when users invite friends, they don’t see or access their friends’ email information directly. Instead, they send a notification request to their friends’ accounts, which ensures the invitation is delivered without exposing the email addresses.
Listing 5.24. Delegated Conglomeration in Java.
1 导入java.util.Vector;
导入java.lang.String;
3
公共类UserAccount {
5 Vector<UserAccount> allMyFriends;
公共字符串 notificationEmail;
7
公共 void invitationToEvent (
9 Vector<UserAccount> friends, String msg) {
for (UserAccount friend : friends) {
11 friend.notify( this , msg);
// ^--- 委托集合
13 }
}
15
公共布尔通知 (UserAccount 邀请者,
17 字符串 msg) {
// 发送电子邮件至notificationEmail
19 return true;
};
21 };
1 import java.util.Vector;
import java.lang.String;
3
public class UserAccount {
5 Vector<UserAccount> allMyFriends;
public String notificationEmail;
7
public void inviteToEvent (
9 Vector<UserAccount> friends, String msg) {
for (UserAccount friend : friends) {
11 friend.notify(this, msg);
// ^--- Delegated Conglomeration
13 }
}
15
public boolean notify (UserAccount inviter,
17 String msg) {
// Send email to notificationEmail
19 return true;
};
21 };
清单 5.23和5.24从技术上讲是委托聚合的实例，但后者的论证更合理，用例更清晰。清单 5.23根据请求者有选择地公开私有数据；清单 5.24表明委托聚合可用于完全隐藏私有数据，使其永远不会被共享。第二个对象被赋予使用私有数据执行任务的唯一责任。
Listings 5.23 and 5.24 are technically instances of Delegated Conglomeration, but the latter shows a much better justification and clearer use case. Listing 5.23 selectively exposes private data based on who requests it; Listing 5.24 shows that Delegated Conglomeration can be used to completely hide private data such that it is never shared. The second object is given the sole responsibility to perform a task with private data.
适用性
在以下情况下使用委托聚合：
Use Delegated Conglomeration when:
• 可以将一项任务分解为多个子任务，每个子任务均可由同一对象类型进行正确处理。
• A task can be broken into subtasks that are properly handled by the same object type.
• 许多同一类型的对象协同工作来完成一项任务。
• Many objects of the same type work in concert to complete a task.
• 单个对象无法单独完成任务。
• A single object cannot complete the task alone.
• 该任务需要其他对象保密的数据。
• The task requires data that is kept private by the other objects.
结构
參與者
委托人
Delegator
包含对其自身类型的其他实例的引用的对象类型。
The object type that contains references to other instances of its own type.
委托目标
delegateTarget
被调用来执行某项任务的封闭实例。
The enclosed instance that is called upon to perform a task.
第一个对象内的调用点。
The calling point within the first object.
操作2
operation2
第二个对象要完成的子任务。
The subtask to be completed by the second object.
合作
此模式仅涉及一种对象类型Delegator，但该类型有两个不同的实例。一个对象依赖另一个对象来执行部分任务（如Delegation ），而该子任务与当前请求没有直接关联（如Conglomeration ） 。
Only one object type, Delegator, is involved in this pattern, but there are two distinct instances of that type. One object relies on the other to perform some part of the task, as with Delegation, and that subtask is not directly associated with the current request, as with Conglomeration.
结果
与重定向递归一样，此模式提供了在多个对象之间分配任务的能力。然而，与其同类不同，委托聚合提供了一个明确的决策点，将行为确定逻辑与行为实现分开。
As with Redirected Recursion, this pattern offers the ability to distribute tasks among a number of objects. Unlike its cousin, however, Delegated Conglomeration offers a clear decision-making point, separating the behavior-determining logic from the behavior implementation.
执行
在 C++ 中：
In C++:
1 class Delegator {
Delegator* delegateTarget;
3 public:
void operation() {
5 // 可选工作优先
delegateTarget->operation2();
7 // 可选工作之后
};
9 void operation2() {
// 做一些有用的事情
11 };
}
1 class Delegator {
Delegator* delegateTarget;
3 public:
void operation() {
5 // Optional work prior
delegateTarget->operation2();
7 // Optional work after
};
9 void operation2() {
// Do something useful
11 };
}
在 Objective-C 中：
In Objective-C:
@interface Delegator
2 {
Delegator* delegateTarget;
4 }
- ( void ) operation;
6 - ( void ) operation2;
@end
8
@implementation Delegator
10 - ( void ) operation
{
12 // [delegateTarget operation2]之前的可选工作；
14 // 之后的可选工作
}
16 - ( void ) operation2
{
18 // 做一些有用的事情
}
20 @end
@interface Delegator
2 {
Delegator* delegateTarget;
4 }
- (void) operation;
6 - (void) operation2;
@end
8
@implementation Delegator
10 - (void) operation
{
12 // Optional work prior
[delegateTarget operation2];
14 // Optional work after
}
16 - (void) operation2
{
18 // Do something useful
}
20 @end
相关模式
显然，委托聚合可以通过放宽对象之间的类型关系转换为委托，也可以通过让对象回调自身转换为聚合。如果将类型相似性更改为子类型，则意味着对象正在调用一组受信任的类，从而产生受信任的委托。保持类型相似性不变并保留不同的对象，但让方法调用另一个对象中的类似对应项来调用相同的行为，这将带我们走向重定向递归。
Obviously, Delegated Conglomeration can be converted either to Delegation, by relaxing the typing relationship between the objects, or to Conglomeration, by having the object call back into itself. If the type similarity is changed to subtyping, it means the object is calling into a family of trusted classes, resulting in Trusted Delegation. Keeping the type similarity the same and retaining the distinct objects, but instead having the method call into its similar counterpart in the other object to invoke the same behavior, brings us to Redirected Recursion.
方法调用分类
对象：不同
Object: Dissimilar
对象类型：类似
Object Type: Similar
方法：不同
Method: Dissimilar
重定向递归
Redirected Recursion
对象行为
Object Behavioral
意图
对多个对象执行单一操作，并让对象负责该操作的分发和调用。
To perform a singular action on multiple objects and have the objects be responsible for distribution and invocation of that action.
动机
我们经常希望对系统中的大量数据执行相同的操作。有时，这些数据最好由外部容器组织，并且这些数据可以被视为“愚蠢的”，不负责自己的组织；只有遍历容器的外部实体才能对其进行操作。遍历容器并对其中包含的每个项目执行操作就是此过程的一个示例。它要求从外部强加数据的结构，并且行为的调用流程也由外部控制。清单 5.25提供了一个简单的示例。
Often we wish to perform the same action on a lot of data in a system. Sometimes that data is best organized by an external container, and the data can be considered “dumb,” with no responsibility for its own organization; it is acted on only by an external entity traversing the container. Iterating over a container and performing an action on each item contained in it is an example of this process. It requires that the structure of the data be imposed from outside and that the flow of invocation of the behavior also be controlled externally. A simple example is provided in Listing 5.25.
Listing 5.25. Traditional iteration and invocation in C.
#include <stdio.h>
2 #define LENGTH 4
int data[LENGTH] = {1, 2, 3, 4};
4 void printAll() {
int i;
6 for (i = 0; i < LENGTH; i++) {
printf("%d\n", data[i]);
8 }
};
#include <stdio.h>
2 #define LENGTH 4
int data[LENGTH] = {1, 2, 3, 4};
4 void printAll() {
int i;
6 for(i = 0; i < LENGTH; i++) {
printf("%d\n", data[i]);
8 }
};
这个解决方案是解决原始问题的常用方法，但可以改进。首先，面向对象系统中的数据不必是愚蠢的。它可以在请求时执行自己的操作。我们可以做一些更像清单 5.26中所示的事情。
This solution is a common way of solving the original problem, but it can be improved on. First, data in an object-oriented system doesn’t have to be dumb. It can perform its own actions when requested. We can instead do something more like what is shown in Listing 5.26.
Listing 5.26. Object-oriented iteration and invocation in C++.
1 #include <cstdio>
3 class DataItem {
int data;
5 public:
DataItem( int val) : data(val) {};
7 void print() {
printf("%d\n", data);
9 };
};
11
#define LENGTH 4
13 DataItem data[LENGTH] =
{DataItem(1), DataItem(2),
15 DataItem(3), DataItem(4)};
17 void printAll() {
int i;
19 for (i = 0; i < LENGTH; i++) {
data[i].print();
21 }
};
1 #include <cstdio>
3 class DataItem {
int data;
5 public:
DataItem(int val) : data(val) {};
7 void print() {
printf("%d\n", data);
9 };
};
11
#define LENGTH 4
13 DataItem data[LENGTH] =
{DataItem(1), DataItem(2),
15 DataItem(3), DataItem(4)};
17 void printAll() {
int i;
19 for(i = 0; i < LENGTH; i++) {
data[i].print();
21 }
};
现在数据负责打印。该printAll()函数不必知道有关数据的任何信息，也不必知道它是如何编码的，甚至不必知道它是如何打印的。它只需要知道如何遍历组织容器并print()在每个对象上调用。
Now the data is responsible for doing the printing. The printAll() function doesn’t have to know anything about the data, how it is encoded, or even how it is printed. It just has to know how to traverse the organizing container and invoke print() on each object.
这个解决方案更好，但我们可以更进一步。数据可能最好用图或平衡树来表示，而不是数组。就目前情况而言，任何希望对上述数组中的每个元素执行操作的代码都必须修改为现在遍历新的组织结构。在大型系统中，这可能是一项艰巨的任务。但是，由于对象可以通过使用Retrieve包含对其他对象的引用，因此我们可以让对象也负责自己的组织，从而得到类似清单 5.27 的内容。
This solution is better, but we can take it a step further. Instead of an array, the data may be best represented by a graph or a balanced tree. As it currently stands, any piece of code that wishes to perform an operation on each element in the above array would have to be modified to now traverse the new organizational structure. In a large system, this can be a huge undertaking. Because objects can contain references to other objects through the use of Retrieve, however, we can have the objects also take care of their own organization, leading us to something like Listing 5.27.
Listing 5.27. Basic Redirected Recursion in C++.
#include <cstdio>
2
class DataItem {
4 int data;
DataItem* next;
6 public:
DataItem( int val) : data(val), next(NULL) {};
8 DataItem( int val, DataItem* next) :
data(val), next(next) {};
10 void print() {
printf("%d\n", data);
12 if (next) {
next->print();
14 // ^--- 重定向递归
}
16 };
};
18
DataItem data =
20 {DataItem(1,
new DataItem(2,
22 new DataItem(3,
new DataItem(4))))};
24
void printAll() {
26 数据.打印();
};
#include <cstdio>
2
class DataItem {
4 int data;
DataItem* next;
6 public:
DataItem(int val) : data(val), next(NULL) {};
8 DataItem(int val, DataItem* next) :
data(val), next(next) {};
10 void print() {
printf("%d\n", data);
12 if (next) {
next->print();
14 // ^--- Redirected Recursion
}
16 };
};
18
DataItem data =
20 {DataItem(1,
new DataItem(2,
22 new DataItem(3,
new DataItem(4))))};
24
void printAll() {
26 data.print();
};
现在，不仅打印行为的责任，而且如何查找下一个数据的责任也已移交给数据对象。由于数据现在负责自己的组织，因此它也可以负责自己的调用顺序。如果类DataItem决定更改其实现并将其伙伴对象存储为红黑树、哈希图或任何其他形式的数据结构，则可以自由地这样做，并且代码printAll()无需更改。这并不总是正确的方法，但在适当的情况下，当系统正在处理的对象可以信任为自组织并且只需要一个触发器即可启动一系列复杂的事件时，这种方法可能非常有用。
Now responsibility not only for the printing behavior but also for how to find the next piece of data has been handed over to the data object. Because the data is now responsible for its own organization, it can also be responsible for its own invocation ordering. If the DataItem class decides to change its implementation and store its partner objects as a red-black tree, a hashmap, or any other manner of data structure, it is free to do so, and the code in printAll() doesn’t need to change. This isn’t always the right approach to take, but it can be powerful in the right circumstances when objects that a system is dealing with can be trusted to be self-organizing and only need a trigger to initiate a complex sequence of events in tandem.
想象一下，一队伞兵准备跳伞。由于空间狭小，指挥官无法走到队伍后面指示每个士兵单独跳伞。相反，他站在队伍的最后面，当跳伞时间到来时，他会拍拍最后一名士兵的肩膀。那名士兵知道拍拍他前面士兵的肩膀，并在那名士兵跳伞后跳伞。这可以继续沿着一条任意长度的队伍继续下去，从 2 名士兵到 200 名士兵。伞兵的唯一任务是，当他们感觉到有人拍他们的肩膀时，拍拍队伍中的下一个人；等待；在有空间时向前移动；当他们看到前面的士兵跳伞时，下一个跳伞。指挥官向每个士兵发出一个命令，而不是一个。此命令的示例代码可能类似于清单 5.28。
Imagine, for example, a line of paratroopers getting ready for a jump. Space is tight, so the commander cannot walk down the line to indicate to each trooper to jump individually. Instead, he stands at the back of the line, and when the drop time comes, he taps the last trooper on the shoulder. That trooper knows to tap the shoulder of the trooper in front of him and to jump after that soldier has jumped. This can continue down a line of arbitrary length, from 2 to 200 troopers. The paratroopers’ only tasks are to, when they feel a tap on their shoulder, tap the next person in line; wait; shuffle forward as space is available; and when they see the soldier in front of them go, jump next. The commander issues one order instead of one to each soldier. A sample coding of this might look like Listing 5.28.
Listing 5.28. Paratroopers implementing Redirected Recursion.
1 class Paratrooper {
bool _hasJumped;
3 Paratrooper* nextTrooper;
public:
5 Paratrooper() : _hasJumped( false ) {};
void jump() {
7 if (nextTrooper) {
nextTrooper->jump();
9 // ^--- 重定向递归
while (! nextTrooper->hasJumped() ) {
11 shuffleForward();
}
13 }
leap();
15 };
void leap() {
17 _hasJumped = true;
// 进入重力的甜蜜怀抱
19 };
void shuffleForward() {
21 // 迈出一步
};
23 bool hasJumped() {
return _hasJumped;
25 };
};
27
class Commander {
29 Paratrooper* backOfLine;
public:
31 void greenLight() {
backOfLine->jump();
33 };
};
1 class Paratrooper {
bool _hasJumped;
3 Paratrooper* nextTrooper;
public:
5 Paratrooper() : _hasJumped(false) {};
void jump() {
7 if (nextTrooper) {
nextTrooper->jump();
9 // ^--- Redirected Recursion
while (! nextTrooper->hasJumped() ) {
11 shuffleForward();
}
13 }
leap();
15 };
void leap() {
17 _hasJumped = true;
// Enter gravity's sweet embrace
19 };
void shuffleForward() {
21 // Take a step
};
23 bool hasJumped() {
return _hasJumped;
25 };
};
27
class Commander {
29 Paratrooper* backOfLine;
public:
31 void greenLight() {
backOfLine->jump();
33 };
};
当前伞兵必须等到前面的伞兵完成任务后才能跳伞，依此类推。最终，每个伞兵都按正确的顺序完成了自己的任务，而且几乎没有任何指令。命令沿着链条向下传递，但行为会反向传递。第一个被击中的伞兵是最后一个跳伞的。
The current paratrooper cannot jump until the trooper in front has completed the task, and so on, and so on. Eventually each paratrooper has completed his or her task, in the proper order, and with very little instruction. The order issues down the chain, but the behavior propagates back. The first trooper tapped is the last to jump.
适用性
在以下情况下使用重定向递归：
Use Redirected Recursion when:
• 递归是将任务分解为子部分的一种简洁方法。
• Recursion is a clean way to break up the task into subparts.
• 同一类型的多个对象必须相互作用才能完成任务。
• Multiple objects of the same type must interact to complete the task.
• 对象可以对自己之间的组织负责。
• The objects can be responsible for their own organization among themselves.
结构
參與者
递归器
Recursor
保存对其自身类型的另一个实例的引用的对象类型。
An object type that holds a reference to another instance of its own type.
重定向目标
redirectTarget
随附的实例。
The enclosed instance.
手术
operation
其中的方法Recursor本身是递归的，但是通过redirectTarget。
A method within Recursor that is recursive on itself but through redirectTarget.
合作
同一对象类型的多个实例交互以完成任务。每个实例都知道如何向下一个对象发送消息以轮到自己。
Multiple instances of the same object type interact to complete a task. Each instance knows how to message the next object for its turn.
结果
这是一种在使用多个对象时实现共享递归行为的强大方法，并且在许多系统中都有使用。它允许将功能拆分到可以自组织的不同数据集中，但它将功能限制为特定行为。它类似于图形处理器等硬件中的单指令、多数据 (SIMD) 计算，只是在这种情况下数据负责传递指令。这可以是一种高度灵活的方法来解决需要分而治之算法的问题，因为更改方法的实现会平等地传播到所有对象。
This is a powerful method for shared recursive behavior when using a number of objects, and it is found in many systems. It allows functionality to be split among disparate data sets that can be self-organized, but it constrains the functionality to a particular behavior. It is analogous to single instruction, multiple data (SIMD) computing in hardware such as a graphics processor, except that in this case the data is responsible for passing along the instruction. This can be a highly flexible approach to solving problems that require a divide-and-conquer algorithm, as changing the method’s implementation propagates across all objects equally.
执行
在 Java 中：
In Java:
public class Recursor {
2 Recursor redirectTarget;
public void operation() {
4 // 之前可选的工作...
redirectTarget.operation();
6 // 之后可选的工作...
};
8 };
public class Recursor {
2 Recursor redirectTarget;
public void operation() {
4 // Optional work before...
redirectTarget.operation();
6 // Optional work after...
};
8 };
在 Python 中：
In Python:
递归类:
2 def __init__(self):
__redirectTarget = Recursor()
4
def operation(self):
6 # 之前的可选工作...
redirectTarget.operation();
8 # 之后的可选工作...
class Recursor:
2 def __init__(self):
__redirectTarget = Recursor()
4
def operation(self):
6 # Optional work before...
redirectTarget.operation();
8 # Optional work after...
相关模式
正如您从名称中可以预料到的那样，通过消除方法相似性要求，重定向递归可以转换为重定向。同样，将不同的对象差异合并为一个对象会导致简单的递归。保留对象差异以加强多个对象之间的合作，同时将方法相似性放宽为方法差异会产生委托聚合。最后，通过将类型关系更改为子类型之一来实现受信任的重定向，这使得方法调用可以通过与发起调用的类相关的受信任类集合以多态方式处理。
As you might expect from the name, Redirected Recursion can be converted into Redirection by eliminating the method similarity requirement. Similarly, collapsing the distinct object dissimilarity into one object results in simple Recursion. Retaining the object dissimilarity to enforce cooperation between multiple objects while relaxing the method similarity to a method dissimilarity yields Delegated Conglomeration. Finally, Trusted Redirection is achieved by changing the type relationship to one of subtyping, which lets the method call be handled polymorphically through a trusted collection of classes related to the one initiating the call.
方法调用分类
对象：不同
Object: Dissimilar
对象类型：类似
Object Type: Similar
方法：类似
Method: Similar
受信任的委托
Trusted Delegation
对象行为
Object Behavioral
意图
相关类通常被定义为集体执行任务。在这些情况下，相关类型的多个对象可以以通用方式交互，以相互委托任务。如果对象属于相关类型，则这些对象可以对它们可以给予另一个对象的信任级别做出某些合理的假设。
Related classes are often defined as such to perform tasks collectively. In these cases, multiple objects of related types can interact in generalized ways to delegate tasks to one another. If the objects are of related types, the objects can make certain well-formed assumptions about the level of trust they can place in the other object.
动机
当对象必须协同行动才能执行任务时（例如在委托中），它们会将一定程度的信任放在另一个对象身上，让后者完成任务。确保信任或至少降低相关风险的一种方法是将调用对象知道的部分任务委托给另一个对象。共享对象信息的一种常见方法是使它们的类型相关，最简单的形式是子类型关系。
When objects must act in concert to perform a task, such as in Delegation, they are placing a certain amount of trust in the other object to complete its task. One way to ensure trust, or at least to reduce the risk involved, is to delegate the portion of the task that the calling object knows something about to another object. A common way to share information about objects is for their types to be related, and the simplest form of that is the subtyping relationship.
通过将任务传递给当前对象的超类型的对象，可以提前了解接收对象的某些方面。方法签名必须在调用实现之前建立，并且可以从方法签名之间的关系推断出意图，就像委托一样。此外，由于调用对象也具有相同的方法，因此可以准确地知道意图。实现可能不同，但调用对象继承了基本概念意图以及签名。它甚至可能具有相同的默认实现。
By passing a task to an object of a supertype of the current object, certain aspects of the receiving object can be known ahead of time. The method signatures are necessarily established before the call can be implemented, and the intent can be assumed from the relationship between the method signatures, as with Delegation. In addition, however, the intent can be known precisely, because the calling object also has that same method. The implementation may differ, but the calling object has inherited the basic conceptual intent as well as the signature. It may even have the same default implementation.
然而，意图和概念在这里是最重要的。调用对象知道委托对象至少将具有超类的语义。通过多态性（如果适用），委托对象可以限制这些语义以提供特殊化，但一般不允许放宽这些语义。调用对象知道委托对象将遵循一组特定的可预先确定的准则。虽然这在任何强类型语言中在技术上都是正确的——因为调用对象必须知道委托对象的类型调用站点 — 类型是调用对象自己的超类型这一事实提供了完全不同级别的知识。这些额外的知识使调用对象能够对委托对象给予更高级别的信任。
Intent and concept, however, are most important here. The calling object knows that the delegate object is going to have at least the semantics of the superclass. Through polymorphism (where applicable), the delegate object may restrict these semantics to provide specialization, but in general, it is not allowed to relax them. The calling object knows that the delegate object will conform to a certain set of predeterminable guidelines. Although this is technically true in any strongly typed language—because the calling object must have an idea of the type of the delegate object at the calling site—the fact that the type is the calling object’s own supertype offers an entirely different level of knowledge. This additional knowledge enables the calling object to place a higher level of trust in the delegate object.
用户界面是一种常见的系统类型，其中可以找到受信任的委托和相关模式，例如代理委托、受信任的重定向和代理重定向。当接口和方法名称已知但精确的对象类型（因此方法主体）可能未知时，此模式允许在类系列（通常称为类簇）内分配任务。它是一种多态委托形式，其中调用对象是多态类型之一。
User interfaces are a familiar type of system in which to find Trusted Delegation and related patterns such as Deputized Delegation, Trusted Redirection, and Deputized Redirection. This pattern allows tasks to be parceled out within a family of classes, often called a class cluster, when the interface and method name are known, but the precise object type and, therefore, method body may not be. It is a form of polymorphic delegation in which the calling object is one of the polymorphic types.
考虑一个窗口系统，其中包括滑块和旋转拨盘作为输入控件，以及文本字段和条形图作为显示窗口小部件。输入控件绑定到特定的显示窗口小部件，并在用户调整控件时向该窗口小部件发送值更新。输入控件不需要确切知道另一端的显示窗口小部件是什么类型；它们只需要知道它们必须updateValue使用适当的值作为参数来调用方法。因为输入控件也必须显示一个值，所以可以通过编程相应地更改它们的调整，所以它们也需要一个updateValue方法。根据我们的继承模式，输入控件和显示窗口小部件似乎属于同一系列，事实上，我们希望确保它们都可以交互，所以我们相应地创建了一个类层次结构，如清单 5.29所示。
Consider a windowing system that includes slider bars and rotary dials as input controls and text fields and bar graphs as display widgets. An input control is tied to a particular display widget and sends that widget updates of values when the control is adjusted by the user. The input controls don’t need to know precisely what kind of display widget is at the other end; they just need to know that they must call the updateValue method with the appropriate value as a parameter. Because input controls also necessarily display a value, it is possible to programmatically change their adjustment accordingly, so they too need an updateValue method. By our Inheritance pattern, it seems the input controls and display widgets are of the same family and, in fact, we want to make sure that they can all interact, so we create a class hierarchy accordingly, as in Listing 5.29.
Listing 5.29. UI widgets demonstrating Trusted Delegation in C++.
class UIWidget {
2 public:
virtual void updateValue( int newValue );
4 };
6 class InputControl : UIWidget{
UIWidget* target;
8 public:
void userHasSetNewValue( int myNewValue) {
10 target->updateValue(myNewValue);
// ^--- 受信任的委托
12 }
};
14
class SliderBar : public InputControl {
16 public:
// 相应地移动滑块
18 void updateValue( int newValue );
void acceptUserClick() {
20 // 确定新值，并设置它
int newVal;
22 // 更新滑块条图形
this ->userHasSetNewValue(newVal);
24 };
};
26
class RotaryKnob : InputControl {
28 public:
// 相应地旋转旋钮图像
30 void updateValue( int newValue );
void acceptUserClick() {
32 // 确定新值，并设置它
int newVal;
34 // 更新旋钮图像
this ->userHasSetNewValue(newVal);
36 };
};
38
class GraphicsContext {
40 public:
virtual void render( int );
42 };
44 class DisplayWidget : public UIWidget {
protected:
46 // 每个子类都应将 // GraphicsContext 设置为对其需求有意义的
内容
48 // GraphicsContext* gc; 50 public: void updateValue( int newValue ) { 52 gc->render( newValue ); }; 54 }; 56类TextWidget :公共DisplayWidget {}; 58类BarGraph :公共DisplayWidget {};
class UIWidget {
2 public:
virtual void updateValue( int newValue );
4 };
6 class InputControl : UIWidget{
UIWidget* target;
8 public:
void userHasSetNewValue(int myNewValue) {
10 target->updateValue(myNewValue);
// ^--- Trusted Delegation
12 }
};
14
class SliderBar : public InputControl {
16 public:
// Moves the slider bar accordingly
18 void updateValue( int newValue );
void acceptUserClick() {
20 // Determine new value, set it
int newVal;
22 // Update the slider bar graphic
this->userHasSetNewValue(newVal);
24 };
};
26
class RotaryKnob : InputControl {
28 public:
// Rotates the knob image accordingly
30 void updateValue( int newValue );
void acceptUserClick() {
32 // Determine new value, set it
int newVal;
34 // Update the knob image
this->userHasSetNewValue(newVal);
36 };
};
38
class GraphicsContext {
40 public:
virtual void render(int);
42 };
44 class DisplayWidget : public UIWidget {
protected:
46 // Each subclass should set the
// GraphicsContext to something
48 // meaningful for its needs
GraphicsContext* gc;
50 public:
void updateValue( int newValue ) {
52 gc->render( newValue );
};
54 };
56 class TextWidget : public DisplayWidget {};
58 class BarGraph : public DisplayWidget {};
在这个例子中，InputControl和类履行了受信任委托UIWidget模式中的必要角色，如图 5.3所示。对于对的调用将做什么有更大的确定性，因为它知道该类，并且通过子类化与该类紧密相关。如果该对象是其他类型，那么对对的调用将完成什么的信息就会少得多。知道它是或其子类之一的实例，表明委托对象将与用户互动，就像它本来的样子。它可以相信其委托任务将按预期得到处理。InputControlupdateValueUIWidgettargetInputControlupdateValue()UIWidgetInputControl
In this example, the InputControl and UIWidget classes fulfill the necessary roles in the Trusted Delegation pattern, as shown in Figure 5.3. InputControl has a greater sense of certainty about what the call to updateValue will do, because it knows about and is intimately tied with the UIWidget class through subclassing. If the target object were of some other type, then InputControl would have much less information about what that call to updateValue() would accomplish. Knowing that it is an instance of UIWidget or of one of its subclasses tells InputControl that the delegate object will be involved with the user, just as it is. It can trust that its delegated task will be handled as expected.
图 5.3. UI 类簇显示了受信任委托的实例。
Figure 5.3. UI class cluster showing an instance of Trusted Delegation.
SliderBar和对象RotaryKnob无需知道它们的值将流向何处，事实上，它们可以相互绑定，彼此同步调整。你甚至可以有两个同一InputControl子类的对象绑定在一起，例如两个SliderBar实例。我们已经将谁发送什么数据和谁接收数据的关注点分开。我们唯一关心的是数据是否以多态方式发送给正确接收的客户端，以及当前调用对象是否属于该多态家族。这允许为许多可以协同工作以执行多项任务的类提供单一、统一的接口。
The SliderBar and RotaryKnob objects do not have to know anything about where their value is going and, in fact, they could be tied to each other, with each adjusting the other in sync. You could even have two objects of the same InputControl subclass tied together, such as two SliderBar instances. We have separated the concerns of who is sending what data and who is receiving it. Our only concerns are that the data is being sent to a properly receiving client polymorphically and that the current calling object is of that polymorphic family. This allows for a single, unified interface for many classes that can work in tandem to perform many tasks.
适用性
在以下情况下使用受信任的委派：
Use Trusted Delegation when:
• 授权是适当的，并执行相关和/或不相关的子任务。
• Delegation is appropriate, with related and/or unrelated subtasks to be performed.
• 任务的要求和/或实施需要一定程度的假定信任。
• A level of assumed trust is required in the requirements and/or implementation for the task.
• 实际实现可能无法提前选择，并且可能需要多态性才能正确处理消息请求。
• The actual implementation may not be selectable ahead of time, and polymorphism may be required to properly handle the message request.
• 调用对象属于多态类层次结构中的一种类型。
• The calling object is of a type in the polymorphic class hierarchy.
结构
參與者
一家之主
FamilyHead
多态类簇的基类。
The base class for a polymorphic class cluster.
委托人
Delegator
的一个子类FamilyHead。
A subclass of FamilyHead.
目标
target
FamilyHead包含的多态实例Delegator。
A polymorphic instance of FamilyHead that is contained by Delegator.
手术
operation
请求受信任的委派任务的调用方法。
The calling method that is requesting a trusted delegated task.
操作2
operation2
被调用的方法被信任执行委托的任务。
The called method trusted to perform the delegated task.
合作
FamilyHead为 提供基础接口Delegator，并且 的实例FamilyHead也驻留在 中以Delegator处理请求。此实例被理解为多态的，并可能以多种不同方式处理请求。受信任委托与受信任重定向的不同之处在于，它是一种更通用的形式，并且将与启动方法相关的子任务（而不是对启动方法的改进）进行分配。
FamilyHead provides a base interface for Delegator, and an instance of FamilyHead also resides within the Delegator to handle requests. This instance is understood to be polymorphic and may handle a request in a number of different ways. Trusted Delegation differs from Trusted Redirection in that it is a more generalized form and parcels out subtasks that are related to, not refinements of, the initiating method.
结果
与任何子类型关系一样，Delegator与的接口相关联FamilyHead。方法的实现target取决于它最终通过多态性包含在其中的特定子类。因此，如果类层次结构中的另一个类以意外的方式实现其方法，则受信任的委托可能会产生意想不到的后果。另一方面，这是一种通过向类系列添加类来扩展功能的强大机制。如果需要以某种方式限制可能的扩展，请考虑使用委托委托模式。
As with any subtyping relationship, Delegator is tied to the interface of FamilyHead. The implementation of the target method is subject to the particular subclass that it ends up being contained in, via polymorphism. For this reason, Trusted Delegation may set up unintended consequences if another class within the class hierarchy implements its methods in an unexpected way. On the other hand, this is a powerful mechanism for extending functionality by adding classes to the class family. If the possible extensions need to be limited in some way, consider using the Deputized Delegation pattern instead.
执行
目标可以在基类或子类中定义；它只需要可以从子类访问。
The target can be defined in either the base class or the subclass; it just needs to be accessible from with the subclass.
In Java, superclass-defined target:
public abstract class FamilyHead {
2 FamilyHead target;
public abstract void operation();
4 public abstract void operation2();
};
6
public class Delegator extends FamilyHead {
8 public void operation() {
// 之前的可选工作...
10 target.operation2();
// ^--- 受信任的委托
12 // 之后的可选工作...
};
14 public void operation2() {};
};
public abstract class FamilyHead {
2 FamilyHead target;
public abstract void operation();
4 public abstract void operation2();
};
6
public class Delegator extends FamilyHead {
8 public void operation() {
// Optional work before...
10 target.operation2();
// ^--- Trusted Delegation
12 // Optional work after...
};
14 public void operation2() {};
};
在 Python 中，子类定义的目标：
In Python, subclass-defined target:
1 class FamilyHead:
def operation(self):
3 pass
def operation2(self):
5 pass
};
7
class Delegator(FamilyHead):
9 target = FamilyHead();
def operation(self) {
11 # 之前的可选工作...
target.operation2();
13 # ^--- 受信任的委托
# 之后的可选工作...
1 class FamilyHead:
def operation(self):
3 pass
def operation2(self):
5 pass
};
7
class Delegator(FamilyHead):
9 target = FamilyHead();
def operation(self) {
11 # Optional work before...
target.operation2();
13 # ^--- Trusted Delegation
# Optional work after...
相关模式
从名称中你可能猜到，受信任的委托是委托的专门版本，但委托将交给受信任类型的对象，因为委托对象的类型是调用对象的超类型。删除此类型关系会导致将此模式变回普通的委托。另一方面，对这种类型关系施加更严格的限制并将其转变为兄弟类型关系会产生代理委托。对类型施加更严格的限制，使其与调用对象完全相同，会创建一个委托聚合的实例。保持类型相似性相同，但修改方法相似性，以便我们在委托对象中调用类似的方法，这将使我们得到受信任的重定向。最后，折叠对象变成同一个，这样我们在同一个对象上调用方法的超类型实现，我们就得到了Revert Method。
As you might guess from the name, Trusted Delegation is a specialized version of Delegation, but the delegation is going to an object of a trusted type, by virtue of that type of the delegate object being a supertype of the calling object. Removing this typing relation results in turning this pattern back into a plain Delegation. On the other hand, putting even tighter restrictions on this typing relationship and turning it into a sibling type relationship yields Deputized Delegation. Putting yet tighter restrictions on the type, making it exactly the same as the calling object, creates an instance of Delegated Conglomeration. Keeping the type smiliarity the same but modifying the method similarity such that we’re calling the similar method in the delegate object gets us to Trusted Redirection. Finally, collapsing the objects into the same one, such that we’re calling the supertype implementation of a method on the same object, we arrive at Revert Method.
方法调用分类
对象：类似
Object: Similar
对象类型：子类型
Object Type: Subtype
方法：不同
Method: Dissimilar
受信任的重定向
Trusted Redirection
对象行为
Object Behavioral
意图
将方法实现的某些部分重定向到当前类所属的可能的类集群。通过类型关系，这些类被视为值得信赖。
Redirect some portion of a method’s implementation to a possible cluster of classes of which the current class is a member. The classes are considered trustworthy through the typing relationships.
动机
通常，在运行时会构建相关对象的分层对象结构，并且需要将行为分布在具有不同实现的多个对象之间。这是将处理请求的责任转移到多个对象之间的有效方法，也是责任链和复合模式中的核心结构 [ 21 ]。可信重定向可以被视为此类链中的单个链接。
Frequently, a hierarchical object structure of related objects is built at runtime, and behavior needs to be distributed among multiple objects operating with differing implementations. This is an effective way to move responsibility for the handling of a request through a number of objects, and it is a core structure in the Chain of Responsibility and Composite patterns [21]. Trusted Redirection can be considered a single link of such chains.
虽然我们可以在这些情况下简单地使用重定向，但相关行为链的要求比简单的方法相似性更为严格。意图必须相同，而不仅仅是相似。通常，甚至实现也被认为具有一些共性。
Although we could simply use Redirection in these cases, chains of related behavior assume more stringent requirements than simple method similarity. The intents must be identical, not just similar. Frequently, even the implementations are assumed to have some commonality.
就像在受信任委派中一样，调用对象和被调用对象之间会建立一定程度的信任，前提是双方都了解被调用对象属于非常密切相关的类型。但是，使用受信任重定向，这种信任会更牢固一些。
Just as in Trusted Delegation, a level of trust is built between the calling object and the called object, based on the knowledge that the called object is of a very closely related type. With Trusted Redirection, however, that trust is just a little bit stronger.
使用重定向意味着调用方法和目标方法背后的意图相似，并且两种方法之间存在很强的相关性。通过确保接收对象的类型是当前对象的超类型，可以建立更强的相关性。声明目标对象必须是当前对象类型的超类或从其派生的类型，可以设置一个受信任类型池。这些类型的行为在形式和意图上应更接近触发调用的方法。
The use of a Redirection implies that the intent behind the calling method and the target method are similar and that a strong correlation exists between the two methods. By ensuring that the type of the recipient object is a supertype of the current object, a much stronger correlation can be established. Declaring that the target object must be of a type that is, or is derived from, a superclass of the current object’s type, sets up a pool of trusted types. These types are expected to have behavior that is closer in form and intent to the method triggering the call.
例如，在事件处理系统中（如 GUI 中），通常有一个事件感知元素的层次结构。数据视图和控件包含在窗格中，窗格组成窗口，窗口位于全局 UI 环境中，可能还有其他元素。任何这些元素中可能有人被要求处理用户生成的事件（例如鼠标单击），但这可能是上下文驱动的处理。例如，在后台且当前处于非活动状态的窗口对鼠标事件的反应可能与在前台的窗口不同。如果窗口处于非活动状态，它将转到前台；但如果窗口处于活动状态，它将把单击事件重定向到其内容，以便它们可以处理它。这些内容可以是选项卡式子窗格、控件小部件、文本框，任何东西。选项卡式子窗格可能同样处于活动状态或非活动状态，并决定如何对单击做出反应。文本框可以根据其状态选择在适当的情况下将单击传递给文本内容。文本内容可以在内部对单击、拖动或其他事件做出反应，同时让文本框处理视觉方面。并非所有事件处理程序都具有视觉组件。所有视觉元素都应能够适当地处理事件，但可能不会要求它们在任何特定时间点以相同的方式处理它们。这种时间动态性是一种常见但并非必要的约束，导致使用可信重定向。
For example, in event-handling systems, such as found in GUIs, it is common to have a hierarchy of event-aware elements. Data views and controls are contained within panes, which are composed into windows, which sit in a global UI environment, perhaps with other elements. Any of these elements may be asked to handle a user-generated event such as a mouse-click, but it may be a context-driven handling. A window that is in the background and currently inactive may react differently to mouse events than one in the foreground, for example. If the window is inactive, it will come to the foreground; but if the window is active, it will redirect the click event to its contents, so they can handle it. Those contents could be a tabbed subpane, a control widget, a text box, anything at all. A tabbed subpane may be similarly active or inactive and decide how to react to the click. A text box may, in turn, elect to pass along the click to the text contents if appropriate, based on its state. The text contents can react to the click, drag, or other event internally while letting the text box handle the visual aspects. Not all event handlers have a visual component. All the visual elements are expected be able to handle an event appropriately, but they may not all be asked to handle them the same way at any specific point in time. This temporal dynamicism is a common but not necessary constraint leading to the use of Trusted Redirection.
事件处理程序实现的示例代码可以用 C++ 定义，如清单 5.30和图 5.4所示。
Example code for an implementation of an event handler could be defined in C++ as in Listing 5.30 and in Figure 5.4.
图 5.4. 显示受信任重定向实例的 UI 类簇。
Figure 5.4. UI class cluster showing an instance of Trusted Redirection.
清单 5.30。C++ 中的事件处理程序显示受信任的重定向。
Listing 5.30. Event handler in C++ showing Trusted Redirection.
#include <vector>
2 using std::vector;
4 class Event {};
class Position {};
6
class MouseEvent : public Event {
8 vector< bool > modifierKeys;
Position position;
10 bool mouseDown;
};
12
class EventHandler {
14 public:
virtual void handle(Event* e) {
16 // 执行需要的操作
};
18 };
20 class TextData : public EventHandler {
// 文本存储类
22 };
24 class UIWidget : public EventHandler {
protected:
26 EventHandler* nextHandler;
public:
28 virtual bool isActive();
virtual void handle(Event* e) {
30 // 可能的设置...
if ( !(this->isActive()) ) {
32 nextHandler->handle(e);
// ^--- 受信任的重定向
34 } else {
// 自己处理
36 }
};
38 };
40 class Button : public UIWidget {
// 任何可点击按钮的基类
42 };
44 class TabPane : public UIWidget {
UIWidget* content;
46 public:
TabPane() {
48 // 设置内容
nextHandler = content;
50 };
};
52
class TextField : public UIWidget {
54 TextData* text;
public:
56 TextField() {
// 设置文本
58 nextHandler = text;
};
60 };
#include <vector>
2 using std::vector;
4 class Event {};
class Position {};
6
class MouseEvent : public Event {
8 vector<bool> modifierKeys;
Position position;
10 bool mouseDown;
};
12
class EventHandler {
14 public:
virtual void handle(Event* e) {
16 // do what is needed
};
18 };
20 class TextData : public EventHandler {
// Text storage class
22 };
24 class UIWidget : public EventHandler {
protected:
26 EventHandler* nextHandler;
public:
28 virtual bool isActive();
virtual void handle(Event* e) {
30 // Possible setup...
if ( !(this->isActive()) ) {
32 nextHandler->handle(e);
// ^--- Trusted Redirection
34 } else {
// Handle it myself
36 }
};
38 };
40 class Button : public UIWidget {
// Base class for any clickable button
42 };
44 class TabPane : public UIWidget {
UIWidget* contents;
46 public:
TabPane() {
48 // Setup contents
nextHandler = contents;
50 };
};
52
class TextField : public UIWidget {
54 TextData* text;
public:
56 TextField() {
// Setup text
58 nextHandler = text;
};
60 };
每个可视化 UI 子类都继承了 的默认行为UIWidget，但可以随时重写。在此结构中，每个子类都使用 的实例EventHandler作为下一个要处理事件的对象，以防当前对象不适合。对此行为的常见更改包括检查传入的事件类型，然后再决定是否处理该事件。
Each visual UI subclass inherits the default behavior from UIWidget, although it can be overridden at any time. In this structure, each subclass uses an instance of EventHandler as the next item to be asked to handle an event in the case that the current object is not suitable. A common alteration to this behavior includes checking to see what type of event has been passed in before making a decision regarding whether to handle it.
适用性
在以下情况下使用受信任的重定向：
Use Trusted Redirection when:
• 预计在编译或运行时组成相关对象的聚合结构。
• An aggregate structure of related objects is expected to be composed at compile or runtime.
• 行为应该分解到各个成员对象。
• Behavior should be decomposed to the various member objects.
• 事先不知道聚合对象的结构。
• The structure of the aggregate objects is not known ahead of time.
• 预计会出现多态行为，但不强制执行。
• Polymorphic behavior is expected but not enforced.
结构
參與者
一家之主
FamilyHead
定义接口并包含可能被覆盖的方法。
Defines interface and contains a method to be possibly overridden.
重定向器
Redirector
使用接口FamilyHead并将内部行为重定向回实例，FamilyHead以获得无定形对象结构的多态行为。
Uses interface of FamilyHead and redirects internal behavior back to an instance of FamilyHead to gain polymorphic behavior over an amorphous object structure.
目标
target
被请求执行重定向工作的对象。
The object that is being requested to perform the redirected work.
指示调用对象和被调用对象中要完成的预期任务。
Indicates the intended task to be completed in both the calling and called objects.
合作
Redirector依赖于类FamilyHead作为接口，并依赖于类的实例作为对象递归实现。重定向关系是此模式的关键部分，因为它推动了“按照要求做同样的事情”的概念。可信方面来自将工作重定向到当前对象的超类型以进行多态处理。
Redirector relies on the class FamilyHead for an interface and an instance of the same for an object-recursive implementation. The Redirection relationship is a critical part of this pattern because it drives the concept of “do the same as I was asked to do.” The trusted aspect comes in from redirecting the work to a supertype of the current object for polymorphic handling.
结果
Redirector依赖于其FamilyHead接口，但整个类层次结构都提供各种实现。因此，层次结构中的其他类可能会表现出意外行为。如果发生意外行为，请考虑使用代理重定向模式将可能性限制在可管理的范围内。
Redirector is reliant on FamilyHead for its interface, but it falls to the entirety of the class hierarchy to provide the various implementations. For this reason, other classes in the hierarchy may exhibit unexpected behavior. If unexpected behavior is occurring, consider instead using the Deputized Redirection pattern to restrict the possibilities to a manageable set.
执行
在 C++ 中：
In C++:
class FamilyHead {
2 public:
virtual void operation();
4 };
6 class Redirector : public FamilyHead {
public:
8 void operation();
FamilyHead* target;
10 };
12 void
Redirector::operation() {
14 // 之前的可选工作...
target->operation();
16 // 之后的可选工作...
}
class FamilyHead {
2 public:
virtual void operation();
4 };
6 class Redirector : public FamilyHead {
public:
8 void operation();
FamilyHead* target;
10 };
12 void
Redirector::operation() {
14 // Optional work before...
target->operation();
16 // Optional work after...
}
相关模式
受信任重定向是重定向的一种特殊化，这样重定向任务的可能目标就来自受信任的类集群，这些类被定义为目标类型的相互子类。因此，放宽这种信任可以让您回到更简单、更通用的 EDP。进一步将这种信任限制为超类的兄弟子类的子集会导致委托重定向。如果需要完全相同类型的目标对象且没有可变性，则请查看重定向递归以了解如何将对象类型相似性设置为相同类型。保留子类型关系但将调用对象和目标对象合并为一个对象会产生扩展方法。只需打破方法相似性即可将我们带入受信任的委托。
Trusted Redirection is a specialization of Redirection such that the possible targets of the redirected task are from a trusted cluster of classes defined as mutual subclasses of a the target type. As such, relaxing that trust lets you go back to that simpler, more general EDP. Further restricting that trust to a subset of the sibling subclasses of the superclass results in Deputized Redirection. If the exact same type of target object is required with no variability, then look to Redirected Recursion for how to work with object type similarity set to the same type. Retaining the subtyping relationship but coalescing the calling and target objects into one object gives rise to Extend Method. Simply breaking the method similarity moves us to Trusted Delegation.
方法调用分类
对象：类似
Object: Similar
对象类型：子类型
Object Type: Subtype
方法：类似
Method: Similar
派出代表团
Deputized Delegation
对象行为
Object Behavioral
意图
当我们需要Trusted Delegation的信任委托行为，但是又觉得它太过笼统时，需要预先选择一个类层次结构的子树，进行受限多态性。
When we need the trusted delegation behavior of Trusted Delegation but find it too generalized, it is necessary to preselect a subtree of the class hierarchy for a restricted polymorphism.
动机
静态类型化是一种从明确定义的池中预选类型并在运行时调用系统之前形成对象类型的更具体概念的方法。这种预选是一种增加类型安全性的方法——我们准确地知道对象在执行期间使用时将具有哪些功能。另一方面，多态性是一种将类型信息抽象出来直到运行时的技术。我们故意混淆对象底层类型的许多细节，以类型安全性为代价提供灵活性。有时我们需要在两者之间取得平衡。
Static typing is a way to preselect types from a well-defined pool and form more concrete notions of an object’s type before the invocation of a system at runtime. This preselection is a way of adding type safety—we know precisely what capabilities an object will have when used during execution. Polymorphism, on the other hand, is a technique for abstracting out typing information until runtime. We purposefully obfuscate many details of the underlying type of the object, providing flexibility at the cost of type safety. Sometimes we need a balance of the two.
考虑来自Trusted Delegation的驱动示例。缺少一个通用的数据输入控件，即用于输入文本的控件。按照设置方式，TextInput从 子类化而来的类InputControl将能够更改BarGraph显示数值的值，而这可能不是设计者想要的。7可以通过添加清单 5.31中所示的类和更改来微调来自Trusted Delegation的设计。
Consider the driving example from Trusted Delegation. A common data-input control is missing, one to enter text. The way things are set up, a TextInput class that subclassed from InputControl would be able to alter the value of a BarGraph showing numeric values, and this may not be what the designer wants.7 The design from Trusted Delegation can be fine-tuned by adding the classes and changes shown in Listing 5.31.
Listing 5.31. UI widgets demonstrating Deputized Delegation in C++.
1 类UIWidget {
公共：
3 虚拟 void updateValue( int newValue );
};
5
类InputControl ：UIWidget{
7 UIWidget* target;
公共：
9 void userHasSetNewValue( int myNewValue) {
target->updateValue(myNewValue);
11 }
};
13
class SliderBar : InputControl {
15 public:
// 相应地移动滑块条
17 void updateValue( int newValue );
void acceptUserClick() {
19 // 确定新值，并设置它
int newVal;
21 // 更新滑块条图形
this ->userHasSetNewValue(newVal);
23 };
};
25
class RotaryKnob : InputControl {
27 public:
// 相应地旋转旋钮图像
29 void updateValue( int newValue );
void acceptUserClick() {
31 // 确定新值，并设置它
int newVal;
33 // 更新旋钮图像
this ->userHasSetNewValue(newVal);
35 };
};
37
class GraphicsContext {
39 public:
virtual void render( int );
41 };
43 class DisplayWidget : public UIWidget {
protected:
45 // 每个子类都应将
// GraphicsContext 设置为对其需求有意义的内容
GraphicsContext *
gc;
49 public:
void updateValue( int newValue ) {
51 gc->render( newValue );
};
53 };
55 class TextWidget : public DisplayWidget {};
57 class BarGraph : public DisplayWidget {};
59 // 支持代理委托的新类
class TextInputControl : public InputControl {
61 TextWidget* textTarget;
public:
63 void userHasSetNewValue( int myNewValue) {
textTarget->updateValue(myNewValue);
65 // ^--- 代理委托
};
67 };
1 class UIWidget {
public:
3 virtual void updateValue( int newValue );
};
5
class InputControl : UIWidget{
7 UIWidget* target;
public:
9 void userHasSetNewValue(int myNewValue) {
target->updateValue(myNewValue);
11 }
};
13
class SliderBar : InputControl {
15 public:
// Moves the slider bar accordingly
17 void updateValue( int newValue );
void acceptUserClick() {
19 // Determine new value, set it
int newVal;
21 // Update the slider bar graphic
this->userHasSetNewValue(newVal);
23 };
};
25
class RotaryKnob : InputControl {
27 public:
// Rotates the knob image accordingly
29 void updateValue( int newValue );
void acceptUserClick() {
31 // Determine new value, set it
int newVal;
33 // Update the knob image
this->userHasSetNewValue(newVal);
35 };
};
37
class GraphicsContext {
39 public:
virtual void render(int);
41 };
43 class DisplayWidget : public UIWidget {
protected:
45 // Each subclass should set the
// GraphicsContext to something
47 // meaningful for its needs
GraphicsContext* gc;
49 public:
void updateValue( int newValue ) {
51 gc->render( newValue );
};
53 };
55 class TextWidget : public DisplayWidget {};
57 class BarGraph : public DisplayWidget {};
59 // New class to support Deputized Delegation
class TextInputControl : public InputControl {
61 TextWidget* textTarget;
public:
63 void userHasSetNewValue(int myNewValue) {
textTarget->updateValue(myNewValue);
65 // ^--- Deputized Delegation
};
67 };
现在，TextInputControl可用于更改TextField类的实例，但无法更改Bar-Graph或其他非文本显示项的实例。SliderBar并且RotaryKnob同样可以限制为基于的类层次结构NumericWidget，例如。这在图 5.5中可能更容易看到。
Now the TextInputControl is available for altering instances of the TextField class, but it will not be able to alter instances of Bar-Graph or other nontext display items. SliderBar and RotaryKnob could similarly be limited to a class hierarchy based on a NumericWidget, for example. This may be easier to see in Figure 5.5.
图 5.5. UI 类集群显示了代理委派的实例。
Figure 5.5. UI class cluster showing an instance of Deputized Delegation.
适用性
在以下情况下使用代理授权：
Use Deputized Delegation when:
•信任委托是适当的通用模式。
• Trusted Delegation is the appropriate general pattern.
• 需要对可能的对象类型进行更好的控制。
• Greater control over the possible types of objects is required.
• 要委托的类型不包括调用对象的类型。
• The types to be delegated to do not include the calling object’s type.
參與者
一家之主
FamilyHead
多态类簇的基类，建立基本的通用接口和语义。
The base class for a polymorphic class cluster, which establishes the basic generalized interface and semantics.
委托人
Delegator
其子类FamilyHead要求将子任务委托给受信任的委托人。
A subclass of FamilyHead that requires a subtask to be delegated to a trusted delegate.
委托兄弟
DelegateSibling
它的另一个子类FamilyHead具有适合子任务的语义。
Another subclass of FamilyHead that has the appropriate semantics for the subtask.
其实例DelegateSibling包含Delegator并将被要求执行子任务。
An instance of DelegateSibling that is contained by Delegator and will be asked to perform the subtask.
手术
operation
调用方法发送要执行的子任务的请求。
The calling method sending the request for the subtask to be performed.
操作2
operation2
被要求执行子任务的调用方法。
The called method being asked to perform the subtask.
结构
合作
除了受信任委托所涉及的协作之外，此模式还添加了DelegateSibling类作为实现的目标operation。通过这样做，它将可能的行为限制为共同祖先类的子类，从而选择精细的语义。
In addition to the collaborations involved in Trusted Delegation, this pattern adds the DelegateSibling class as the target for the operation implementation. By doing so, it limits the possible behavior to a subclass of the common ancestor class and therefore to select refined semantics.
结果
此模式与其同类模式“受信任委托”不同，设计决定将多态性静态地限制在原始类系列的子树中。它可以在控制复杂性方面提供很大的好处，但也可能导致以后出现问题，例如，如果所选的兄弟类是被认为过于局限。尽管对代码的语法更改很小，但添加更广泛的可能类和行为可能会产生深远的影响，应仔细考虑。还请注意，与受信任的委托不同，target不能成为共享超类的成员，因为超类必须事先了解其子类之一。虽然有些语言可以实现这种特殊的技巧，但这不是一个健全的设计决策，应尽可能避免。
This pattern differs from its cousin, Trusted Delegation, in that a design decision is made to limit the polymorphism statically to a subtree of the original class family. It can provide large benefits in containing complexity, but it can also lead to issues later if, for example, the chosen sibling class is determined to be too limiting. Although the syntactic change to the code is minimal, the addition of a broader set of possible classes and behaviors can have far-reaching effects that should be carefully considered. Also note that unlike Trusted Delegation, the target cannot be a member of the shared superclass, because the superclass would have to have prior knowledge of one of its subclasses. Although some languages can pull off this particular bit of trickery, it’s not a robust design decision and should be avoided whenever possible.
执行
在 C++ 中：
In C++:
1 class FamilyHead {
protected:
3 void operation();
void operation2();
5 };
7 class DelegateSibling : public FamilyHead {
void operation2();
9 };
11 class Delegator : public FamilyHead {
DelegateSibling* target;
13 void operation() {
// 可选的工作优先
15 target->operation2();
// ^--- 代理委托
17 // 可选的工作优先
};
19 };
1 class FamilyHead {
protected:
3 void operation();
void operation2();
5 };
7 class DelegateSibling : public FamilyHead {
void operation2();
9 };
11 class Delegator : public FamilyHead {
DelegateSibling* target;
13 void operation() {
// Optional work prior
15 target->operation2();
// ^--- Deputized Delegation
17 // Optional work after
};
19 };
相关模式
代理委托相当专业，只能导致几个更通用的 EDP。如果使用可信类簇的标准稍微放宽，将对象类型相似性设置为子类型，我们就会回到可信委托。如果被调用的方法更改为与调用者相似的方法，那么我们就会转向代理重定向。在这种情况下，将调用和委托对象合并为一个没有多大意义，最终会得到格式错误的 EDP 设计空间位置。
Deputized Delegation is rather specialized and can only lead to a couple of more general EDPs. If the criteria of using a trusted class cluster is relaxed slightly such that the object type similarity is set to subtyping, we get back to Trusted Delegation. If the method being invoked is changed to one that is similar to the caller, then we slide over to Deputized Redirection. This is one case where collapsing the calling and delegate objects into one doesn’t make much sense, and we end up with ill-formed EDP design space positions.
方法调用分类
对象：不同
Object: Dissimilar
对象类型：兄弟
Object Type: Sibling
方法：不同
Method: Dissimilar
代理重定向
Deputized Redirection
对象行为
Object Behavioral
意图
当我们需要受信任重定向的受信任重定向行为但发现它仍然过于通用时，需要预先选择类层次结构的子树以实现受限多态性。所选的类子树将不包括发起重定向的类。
When we need the trusted redirection behavior of Trusted Redirection but find it still too generalized, it is necessary to preselect a subtree of the class hierarchy for a restricted polymorphism. This selected subtree of classes will not include the class originating the redirection.
动机
静态类型化是一种从明确定义的池中预选类型并在运行时调用系统之前形成对象类型的更具体概念的方法。这种预选是一种增加类型安全性的方法——我们准确地知道对象在执行期间使用时将具有哪些功能。另一方面，多态性是一种将类型信息抽象出来直到运行时的技术。我们故意混淆对象底层类型的许多细节，以类型安全性为代价提供灵活性。有时我们需要在两者之间取得平衡。
Static typing is a way to preselect types from a well-defined pool and form more concrete notions of an object’s type before the invocation of a system at runtime. This preselection is a way of adding type safety—we know precisely what capabilities an object will have when used during execution. Polymorphism, on the other hand, is a technique for abstracting out typing information until runtime. We purposefully obfuscate many details of the underlying type of the object, providing flexibility at the cost of type safety. Sometimes we need a balance of the two.
在Trusted Redirection的驱动示例中，代码提供了一个非常通用的事件处理系统。然而，这种开放式类型层次结构并不适用于所有情况。例如，滑块只会响应鼠标事件，除非在特殊情况下，否则不会响应击键。同样，文本字段将响应鼠标事件和击键，但击键的处理可能最好由底层文本数据对象来处理。因此，通过简单的改进，可以在运行时消除对整个事件类的考虑。从Trusted Redirection中的清单 5.30中的代码开始，我们可以将其更改为清单 5.32中的代码。
In the driving example for Trusted Redirection, the code provided a very general event-handling system. Such an open-ended type hierarchy, however, is not appropriate for all cases. For instance, a slider bar is only going to respond to mouse events, and not, except in special circumstances, to keystrokes. Likewise, a text field will respond to both mouse events and keystrokes, but the handling of keystrokes may be best taken care of by an underlying text data object. Entire classes of events can therefore be eliminated from consideration at runtime by a simple refinement. Starting with the code in Listing 5.30 from Trusted Redirection, we can alter it to look like the code in Listing 5.32.
Listing 5.32. UI widgets demonstrating Deputized Redirection in C++.
1 #include <vector>
使用std::vector;
3
类Position {};
5 类Key {};
类Event {
7 公共：
virtual int getID();
9 };
11 class MouseEvent : public Event {
vector< bool > modifierKeys;
13 Position position;
bool mouseDown;
15 };
17 class KeyEvent : public Event {
vector< bool > modifierKeys;
19 Key key;
bool keyDown;
21 };
23 class EventHandler {
public:
25 virtual void handle(Event* e) {
// 执行需要执行的操作
27 };
};
29
class TextData : public EventHandler {
31 public:
// 文本存储类
33 virtual void handle(KeyEvent* e) {
// 执行需要执行的操作
35 };
};
37
class UIWidget : public EventHandler {
39 protected:
EventHandler* nextHandler;
41 public:
bool isActive();
43 virtual void handle(Event* e) {
// 可能的设置...
45 if ( !( this ->isActive()) ) {
nextHandler->handle(e);
47 } else {
// 自己处理
49 }
};
51 };
53 class Button : public UIWidget {
// 任何可点击按钮的基类
55 };
57 class TabPane : public UIWidget {
UIWidget* content;
59 public:
TabPane() {
61 // 设置内容
nextHandler = content;
63 };
};
65
class TextField : public UIWidget {
67 TextData* text;
public:
69 TextField() {
// 设置文本
71 nextHandler = text;
};
73 virtual void handle(Event* e) {
// ea 是 KeyEvent 还是子类？
75 if ( dynamic_cast <KeyEvent*>(e)) {
// 之前可选的工作
77 text->handle( dynamic_cast <KeyEvent*>(e));
// ^--- 代理重定向
79 // 之后可选的工作
} else {
81 // 处理 MouseEvents
}
83 };
};
1 #include <vector>
using std::vector;
3
class Position {};
5 class Key {};
class Event {
7 public:
virtual int getID();
9 };
11 class MouseEvent : public Event {
vector<bool> modifierKeys;
13 Position position;
bool mouseDown;
15 };
17 class KeyEvent : public Event {
vector<bool> modifierKeys;
19 Key key;
bool keyDown;
21 };
23 class EventHandler {
public:
25 virtual void handle(Event* e) {
// do what is needed
27 };
};
29
class TextData : public EventHandler {
31 public:
// Text storage class
33 virtual void handle(KeyEvent* e) {
// do what is needed
35 };
};
37
class UIWidget : public EventHandler {
39 protected:
EventHandler* nextHandler;
41 public:
bool isActive();
43 virtual void handle(Event* e) {
// Possible setup...
45 if ( !(this->isActive()) ) {
nextHandler->handle(e);
47 } else {
// Handle it myself
49 }
};
51 };
53 class Button : public UIWidget {
// Base class for any clickable button
55 };
57 class TabPane : public UIWidget {
UIWidget* contents;
59 public:
TabPane() {
61 // Setup contents
nextHandler = contents;
63 };
};
65
class TextField : public UIWidget {
67 TextData* text;
public:
69 TextField() {
// Setup text
71 nextHandler = text;
};
73 virtual void handle(Event* e) {
// Is e a KeyEvent or subclass?
75 if (dynamic_cast<KeyEvent*>(e)) {
// Optional work before
77 text->handle(dynamic_cast<KeyEvent*>(e));
// ^--- Deputized Redirection
79 // Optional work after
} else {
81 // Handle MouseEvents
}
83 };
};
新的结构如图 5.6所示。我们添加了KeyEvent类，TextField现在覆盖了handle()。它MouseEvent自己处理了的实例，但以前它只是将设置next-Handler为TextData实例text。有可能 aMouseEvent会通过错误溜走并绊倒TextData的实现handle()，但这样传递给的唯一事件TextData就是 的事件KeyEvent。TextField和TextData的信任级别高于 doTextField和 的其他子类Event-Handler。
The new structure is shown in Figure 5.6. We added the KeyEvent class, and TextField now overrides handle(). It handles instances of MouseEvent on its own, but previously it simply set the next-Handler to the TextData instance text. There was the chance that a MouseEvent could slip through via a bug and trip up TextData’s implementation of handle(), but this way the only events handed to TextData are those of KeyEvent. TextField and TextData have a higher level of trust than do TextField and other subclasses of Event-Handler.
图 5.6。UI 类集群显示了代理重定向的实例。
Figure 5.6. UI class cluster showing an instance of Deputized Redirection.
现在，TextField确保它处理，MouseEvents而底层TextData仅处理原始文本交互。TextData值得信赖来处理KeyEvent事件，因为TextField它更准确地知道将如何TextData处理它们。此外，TextField它明确不将自己作为可能的接收者的类型，就像在受信任重定向中的情况一样。TextField它坚持要求与之相关的其他人来处理这部分工作。
Now, TextField ensures that it handles the MouseEvents while the underlying TextData only handles raw text interactions. TextData is trusted to handle the KeyEvent events, because TextField knows more precisely what to expect TextData will do with them. In addition, TextField is specifically not including itself as the type of the possible recipient, as is the case in Trusted Redirection. TextField is insisting that someone else it is related to take care of that portion of the job.
适用性
在以下情况下使用代理重定向：
Use Deputized Redirection when:
•可信重定向是适当的通用模式。
• Trusted Redirection is the appropriate general pattern.
• 需要对可能的对象类型进行更好的控制。
• Greater control over the possible types of objects is required.
• 要重定向到的类型不包括调用对象的类型。
• The types to be redirected to do not include the calling object’s type.
參與者
一家之主
FamilyHead
定义接口，包含可能被重写的方法，并且是和的基Redirector类RedirectSibling。
Defines interface, contains a method possibly to be overridden, and is the base class for both Redirector and RedirectSibling.
重定向器
Redirector
使用接口FamilyHead；将内部行为重定向回在无定形但范围有限的RedirectSibling对象结构上获得多态行为的一个实例。
Uses interface of FamilyHead; redirects internal behavior back to an instance of RedirectSibling to gain polymorphic behavior over an amorphous but limited-in-scope object structure.
重定向兄弟
RedirectSibling
多态行为的新类树的头部。
The head of a new class tree for polymorphic behavior.
目标
target
RedirectSibling包含的多态实例Redirector。
A polymorphic instance of RedirectSibling that is contained by Redirector.
手术
operation
呼叫站点。
The calling site.
结构
合作
与受信任重定向一样，Redirector和FamilyHead通过接口和意图绑定。RedirectSibling为特定类型的实现提供了概念起点。在这种情况下，Redi-rector具体将可能成为请求多态目标的类集缩减为的类或子类。和在中具有共同的基类Redirect-Sibling这一事实意味着彼此可以对彼此做出更强的假设。RedirectorRedirectSiblingFamilyHead
As with Trusted Redirection, Redirector and FamilyHead are tied by interface and intent. The RedirectSibling provides a conceptual starting point for a particular type of implementation. In this case, Redi-rector specifically reduces the set of possible classes that can be polymorphic targets of the request to either classes or subclasses of Redirect-Sibling. The fact that Redirector and RedirectSibling have a common base class in FamilyHead means that each may make stronger assumptions about the other.
结果
此模式与其同类模式“受信任重定向”的不同之处在于，设计决定将多态性静态地限制在原始类系列的子树中。它可以在控制复杂性方面提供巨大优势，但如果确定所选的同级类限制过多，则以后也可能导致问题。虽然对代码的语法更改很小，但添加更广泛的可能类和行为可能会产生深远的影响，应仔细考虑。
This pattern differs from its cousin Trusted Redirection in that a design decision has been made to limit the polymorphism statically to a subtree of the original class family. It can provide large benefits in containing complexity, but it can also lead to issues later if, for example, the chosen sibling class is determined to be too limiting. Although the syntactic change to the code is minimal, the addition of a broader set of possible classes and behaviors can have far-reaching effects that should be carefully considered.
执行
在 C++ 中：
In C++:
class FamilyHead {
2 protected:
virtual void operation();
4 };
6 class RedirectSibling : public FamilyHead {
void operation();
8 }
10 class Redirector : public FamilyHead {
public:
12 RedirectSibling* target;
void operation() {
14 // 可选工作，优先
target->operation();
16 // ^--- 代理重定向
// 可选工作，之后
18 };
};
class FamilyHead {
2 protected:
virtual void operation();
4 };
6 class RedirectSibling : public FamilyHead {
void operation();
8 }
10 class Redirector : public FamilyHead {
public:
12 RedirectSibling* target;
void operation() {
14 // Optional work prior
target->operation();
16 // ^--- Deputized Redirection
// Optional work after
18 };
};
相关模式
因为委托重定向是受信任重定向的特例，如果我们通过将类型相似性从兄弟关系改回更一般的子类型关系来放松信任，我们就可以返回到该 EDP。或者，我们可以更改方法相似性以调用不同的方法并得出委托代理。最后，应该注意的是，在这种情况下试图将对象相似性更改为相似是没有意义的，我们得出了 EDP 设计空间中未定义的坐标之一。
Because Deputized Redirection is a specialization of Trusted Redirection, we can return to that EDP if we relax the trust a bit by changing the type similarity from a sibling relationship back to the more general subtyping relationship. Or, we can change the method similarity to call a dissimilar method and arrive at Deputized Delegation. Finally, it should be noted that attempting to change the object similarity to be similar makes no sense in this case, and we arrive at one of our undefined coordinates in the EDP design space.
方法调用分类
对象：不同
Object: Dissimilar
对象类型：兄弟
Object Type: Sibling
方法：类似
Method: Similar
6. 中间图案组合
6. Intermediate Pattern Compositions
本章建立了直接由 EDP 组成的第一级模式。我们在第 4 章中看到了这个过程，在这里我们充实了几个常见、简单但尚未成为基本设计模式的模式。与以前一样，第 3 章中的模式实例符号 (PIN)用于说明构成这些模式的组件部分。对于 EDP，PIN 符号用于使用 UML 显示内部关系。从现在开始，UML 和 PINbox 都出现在 PIN 图中。还包括一个仅限 PINbox 的版本，以说明纯概念连接。
This chapter establishes the first level of patterns composed directly from EDPs. We saw this process in Chapter 4, and here we flesh out several common, simple, but not-yet-elemental design patterns. As before, the Pattern Instance Notation (PIN) from Chapter 3 is used to illustrate the component portions that form these patterns. With the EDPs, the PIN notation was used to show the internal relationships using UML. From here on out, both UML and PINboxes appear in the PIN diagrams. A PINbox-only version is also included to illustrate the purely conceptual connections.
这些模式非常常见，开发人员不断回顾它们，因为它们不仅有效，而且还以新颖的方式组合了少量简单概念。中间模式不是 EDP，因为它们是可分解的，可以作为更基本概念的组成来讨论。然而，在讨论这些模式时，底层概念的交集通常不会明确说明。学生或开发人员必须自己去领悟模式的深层语义。这假设模式的读者对底层概念有足够的练习，能够在出现时识别它们。
These patterns are extremely common, and developers keep coming back to them because they are both effective and because they also combine a small number of simple concepts in novel ways. The intermediate patterns are not EDPs because they are decomposable and can be discussed as the composition of more fundamental concepts. Frequently, however, the intersection of the underlying concepts is not made explicit in discussions of these patterns. It is left up to the student or developer to have that eureka moment of comprehension into the deeper semantics of the pattern. This assumes that the reader of the pattern has sufficient practice with the underlying concepts to recognize them when they occur.
使用 EDP 作为基础，我们可以使构成每个模式的概念的组成和连接点更加明确。本章向您展示了如何快速构建小概念以形成更复杂的抽象，并演示了如何分析已发布的模式以获得新的见解。研究这些中间模式的内部工作原理将为您理解如何结合使用 EDP 和其他模式来形成更大、更丰富的设计模式奠定坚实的基础。
Using the EDPs as a basis, we can make the composition and connection points of the concepts that form each pattern much more explicit. This chapter shows you how small concepts can quickly build into more complex abstractions and demonstrates how to analyze already published patterns for new insights. Studying the inner workings of these intermediate patterns will give you a strong foundation for understanding how EDPs and other patterns can be used in conjunction to form larger and richer design patterns.
本章中的模式规范不包含方法调用分类部分，因为该部分特定于方法调用 EDPS。
The pattern specifications in this chapter do not contain the method-call Classification section, as that was specific to the method-call EDPS.
履行方法
Fulfill Method
对象结构
Object Structural
意图
为先前抽象的方法提供实现，从而履行抽象接口所制定的契约，即以后的子类将提供这样的实现。
To provide an implementation for a previously abstracted method, thereby fulfilling the contract made by Abstract Interface that a later subclass would provide such an implementation.
动机
抽象接口承诺某个子类将为旨在成为超类的类中声明的方法接口提供适当的方法实现。继承让我们可以定义该子类，而实现方法则描述了如何通过结合两个模式实例来实现先前的承诺。
Abstract Interface promises that some subclass will provide a proper method implementation to a method interface declared in a class intended to be a superclass. Inheritance lets us define that subclass, and Fulfill Method describes how to go about fulfilling that prior promise by combining the two pattern instances.
这种方法使我们能够将超类中方法的抽象与可能（或不）提供其实现的任意数量的子类完全分开。
This approach lets us cleanly separate the abstraction of the method in the superclass from any number of possible subclasses that may (or may not) provide an implementation for it.
适用性
在以下情况下使用Fulfill 方法：
Use Fulfill Method when:
• 使用抽象接口推迟了方法的实现。
• Implementation of a method has been deferred using Abstract Interface.
•抽象接口中指定的类的子类能够按照承诺处理方法的功能。
• A subclass of the class specified in Abstract Interface is able to handle the functionality of the method as promised.
參與者
摘要者
Abstractor
为 声明接口的类类型operation。
The class type that declares an interface for operation.
履行者
Fulfiller
定义方法体的类类型operation；继承自Abstractor。
The class type that defines a method body for operation; inherits from Abstractor.
手术
operation
Abstractor在 中声明并定义的方法Fulfiller。
The method that was declared in Abstractor and defined in Fulfiller.
结构
PINbox 专用版本（也用于图 4.4）：
PINbox-only version (also used in Figure 4.4):
合作
Abstractor为方法定义一个接口，并Fulfiller提供实现。Fulfiller从子类中Abstractor获得适当的类型和接口。
Abstractor defines an interface for a method, and Fulfiller provides the implementation. Fulfiller subclasses from Abstractor to gain the appropriate typing and interface.
Abstractor是抽象接口实例中的主要类，并且Abstractor和通过继承Fulfiller实例相关。
Abstractor is the primary class in the instance of Abstract Interface, and Abstractor and Fulfiller are related through an instance of Inheritance.
结果
Fulfiller依赖于Abstractor实现方法的接口。这两者必须保持同步，有时类层次结构顶部的混洗接口会对其子类产生深远的影响。在预期这种情况的情况下，可能首选使用委托的解决方案。使用委托的解决方案允许外部接口独立于实现对象。或者，如果只要求一个类/对象，则可以使用聚合将任务移交给类型内部的私有方法，从而保留接口/实现封装。
Fulfiller relies on Abstractor for the interface to the implemented method. These two must be kept in sync, and there are times when a shuffling interface at the top of a class hierarchy can have far-reaching effects on its subclasses. In cases where this is expected, a solution utilizing Delegation may be preferred. A solution using Delegation allows the external interface to be independent of the implementation object. Alternatively, if only one class/object is mandated, Conglomeration can be used to hand off the task to private methods internal to the type, thereby preserving the interface/implementation encapsulation.
执行
任何支持继承和抽象方法的语言，无论其表达方式如何，都支持Fulfill Method。以下是基于Abstract Interface中的示例的一些示例。
Every language that supports inheritance and abstract methods, regardless of expression, supports Fulfill Method. Following are some examples based on the examples from Abstract Interface.
在 C++ 中：
In C++:
1 class Abstractor {
public:
3 virtual void operation() = 0;
};
5
class Fulfiller :
7 public Abstractor {
public:
9 void operation() {
// 执行适当的工作
11 };
};
1 class Abstractor {
public:
3 virtual void operation() = 0;
};
5
class Fulfiller :
7 public Abstractor {
public:
9 void operation() {
// Perform the appropriate work
11 };
};
在 Java 中，该方法包含在接口中或位于抽象类中：
In Java, the method is included in an interface or is in an abstract class:
public interface Abstractor {
2 public void operation();
};
4
public abstract class AnotherAbstractor {
6 public abstract void operation2();
public void operation3();
8 };
10 public class Fulfiller
extends AnotherAbstractor
12 implements Abstractor {
public void operation() {
14 // 执行适当的工作
}
16 public void operation2() {
// 执行适当的工作
18 }
};
public interface Abstractor {
2 public void operation();
};
4
public abstract class AnotherAbstractor {
6 public abstract void operation2();
public void operation3();
8 };
10 public class Fulfiller
extends AnotherAbstractor
12 implements Abstractor {
public void operation() {
14 // Perform the appropriate work
}
16 public void operation2() {
// Perform the appropriate work
18 }
};
在 Python 3.x 中：
In Python 3.x:
1 class Abstractor(metaclass=ABCMeta):
@abstractmethod
3 def operation(self, ...):
// 允许默认实现
5 return
7 class Fulfiller(Abstractor):
def operation(self, ...):
9 // 执行适当的工作
传递
1 class Abstractor(metaclass=ABCMeta):
@abstractmethod
3 def operation(self, ...):
// Default implementation allowed
5 return
7 class Fulfiller(Abstractor):
def operation(self, ...):
9 // Perform the appropriate work
pass
相关模式
Fulfill Method由两个核心 EDP 组成：抽象接口和继承。由于Fulfill Method实现了抽象接口的承诺，因此几乎在有抽象接口的所有情况下都会出现它。在Fulfill Method不伴随抽象接口的情况下，生成的抽象类在代码中不可用，直到子类完成承诺。本章后面的Objectifier下可以找到多个Fulfill Method实例。Fulfill Method也是大多数四人帮模式的必要组成部分，包括Proxy、Command、Iterator、Observer、State、Decorator、Prototype、Template Method和许多其他模式 [ 21 ]。
Fulfill Method is composed of two of the core EDPs: Abstract Interface and Inheritance. Because Fulfill Method fulfills the promise made by Abstract Interface, it is found in almost every case where Abstract Interface is. In those cases where Fulfill Method doesn’t accompany Abstract Interface, the resulting abstract class isn’t usable in code until a subclass completes the promise. Multiple Fulfill Method instances can be found later in this chapter under Objectifier. Fulfill Method can also be found as a necessary component of most of the Gang of Four patterns, including Proxy, Command, Iterator, Observer, State, Decorator, Prototype, Template Method, and many others [21].
检索新
Retrieve New
对象管理
Object Management
意图
当需要新实例并且该实例的创建过程复杂或昂贵且需要封装在另一个对象中时使用。保证返回的对象没有任何其他引用。
Used when a new instance is needed and the creation of that instance is a complex or expensive procedure that needs to be encapsulated in another object. The returned object is guaranteed not to have any other references to it.
动机
通常，新创建的对象需要是具有明确所有权的原始对象。本地创建这样的对象始终是一种选择，但如果要创建的对象是由相当复杂的行为决定的，那么它可能不是一个最佳的决定。一个例子是根据对数据库查询的响应给出的答复来决定创建哪个对象。任何包含在复杂逻辑或可能需要大量时间的事务中的对象创建过程都不是适合复制和粘贴重用的代码。我们希望既集中这种逻辑，又将其封装到自己的方法中。
Often, a newly created object needs to be a pristine object with well-defined ownership. Creating such an object locally is always an option, but it can be a suboptimal decision if, for instance, the object to be created is determined by a rather complex behavior. An example would be deciding which object to create on the basis of which reply is given in response to a database query. Any object creation process that is wrapped in complex logic or transactions that may take significant time isn’t code that lends itself to copy and paste reuse. We want to both centralize such logic and encapsulate it into its own method.
当然，这种逻辑封装会导致其他问题。如果创建的对象可以被多个来源引用，并且内存模型是非托管的，例如在 C++ 中，那么任何引用该对象的代码都可以请求销毁它，而其余的引用将指向无效对象。或者，没有任何引用对象可以请求销毁它，从而导致内存泄漏，因为对象在仍然驻留时变得无法访问。垃圾收集可以帮助解决这个问题，但定义良好的内存管理始于小决策。
This encapsulation of logic leads to other issues, of course. If the created object can be referenced by a number of sources and the memory model is an unmanaged one such as in C++, then any piece of code with such a reference to the object can request its destruction, leaving the remainder of the references pointing to an invalid object. Alternatively, none of the referring objects may request its destruction, leading to a memory leak because the object becomes unreachable while still resident. Garbage collection can help with this issue, but well-defined memory management starts with small decisions.
其中一个决定是确保只有Retrieve的接收者才能访问被检索的对象，从而建立明确而精确的所有权。这可以通过返回对象的副本几乎透明地完成，因为副本将创建一个新的实例。其他方案涉及传回被销毁的方法本地引用可以在方法退出时使用，但除非对所涉及的所有权问题有强烈的意识和执行，否则它们在以后的维护过程中更容易出错。在所有情况下，这也需要应用Create Object的实例。
One such decision is to make sure that only the recipient of a Retrieve has access to the object being retrieved, thereby establishing a clear and precise ownership. This can be done almost transparently by returning a copy of an object, because the copy will create a fresh instance. Other schemes involving passing back a method-local reference that is destroyed on method exit can be used, but they are more error prone during later maintenance unless there is a strong sense for, and enforcement of, the ownership issues involved. In all cases, this requires an instance of Create Object to be applied as well.
适用性
在以下情况下使用“检索新内容”：
Use Retrieve New when:
• 远程对象提供本地计算所需的对象，由公开的字段对象或公开方法的返回值提供。
• A remote object provides an object that is required for local computation and is provided by an exposed field object or return value from an exposed method.
• 返回的对象需要是一份全新的副本，且不能有其他对它的引用。
• The object returned needs to be a fresh copy without other references to it.
结构
PINbox-only version (identical to Retrieve Shared):
參與者
来源
Source
负责交出所检索对象的对象类型。它充当初始源，创建对象。
The type of the object tasked with handing off the retrieved object. It serves as the initial source, creating the object.
下沉
Sink
target请求检索的对象并包括要赋予新值的项目的对象（或类）类型。
The object (or class) type that requests the retrieved object and includes the item, target, to be given a new value.
检索
Retrieved
要更新的值的类型和返回的值。
The type of the value to be updated and the value that is returned.
目标
target
更新到检索到的对象以供本地使用的字段。
The field that is updated to the retrieved object for local use.
已选定
selected
生成并返回新值的方法或字段。在这种情况下，保证返回的对象没有任何其他引用。
The method or field that produces and returns the new value. In this scenario, the returned object is guaranteed to not have any other references to it.
合作
这个协作中只有三个对象，它们只是扮演了请求发起者、请求履行者和传递对象的角色。Retrieve New的关键部分发生在检索方法的实现中selected。此方法的任务是确保返回的对象是新形成的，并且没有其他对它的引用。
There are only three objects in this collaboration, and they simply play the parts of the request originator, the request fulfiller, and the passed object. The critical portions of Retrieve New occur within the implementation of the retrieval method selected. This method is tasked with ensuring that the object being returned is newly formed and has no other references to it.
结果
对象所有权与对象创建之间的分离具有一些优势。例如，可以灵活地确定所创建对象的类型，可能使用多态性，例如在抽象工厂中。然而，这意味着接收者对象对所创建对象的生命周期负全部责任。如果分发了对所创建对象的引用，则必须制定良好的引用管理策略，除非允许垃圾回收。
Separation between object ownership and object creation has some advantages. For example, the type of the created object can be determined in a flexible manner, possibly using polymorphism, such as in an Abstract Factory. It means, however, that the recipient object is solely responsible for the lifetime of the created object. If references to the created object are handed out, a good reference management policy must be in place unless garbage collection is allowed.
执行
在 C++ 中：
In C++:
类Retrieved {};
2
类Source {
4 公共：
Retrieved giveMeAValue() {
6 检索 ret;
返回ret;
8 };
};
10
类Sink {
12 检索目标;
源 srcobj;
14 公共：
void operation() {
16 target = srcobj.giveMeAValue();
;}
18 };
class Retrieved {};
2
class Source {
4 public:
Retrieved giveMeAValue() {
6 Retrieved ret;
return ret;
8 };
};
10
class Sink {
12 Retrieved target;
Source srcobj;
14 public:
void operation() {
16 target = srcobj.giveMeAValue();
;}
18 };
在 Python 中：
In Python:
1 类源：
def giveMeAValue（self）：
3 ret = Retrieved（）
返回ret
5
类接收器：
7 def __init__（self）：
srcobj = Source（）
9 返回自身
11 def操作（自身）：
目标 = srcobj.giveMeAValue（）
1 class Source:
def giveMeAValue(self):
3 ret = Retrieved()
return ret
5
class Sink:
7 def __init__(self):
srcobj = Source()
9 return self
11 def operation(self):
target = srcobj.giveMeAValue()
相关模式
检索新对象与检索共享对象密切相关，如果您查看 PINbox 专用图，您会发现它们是等效的。不同之处在于行为层面，因为请求的对象返回给调用者的方式不同，要么像这里一样具有清晰的所有权特征，要么像在检索共享对象中一样保留可能的引用。
Retrieved New is closely related to Retrieve Shared, and if you look at the the PINbox-only diagrams, they are equivalent. The distinctive portion is a behavior-level aspect because of how the requested object is returned to the caller, either with clean ownership characteristics, as here, or with possible references being retained, as in Retrieve Shared.
原始设计模式[ 21 ]中的许多创建模式都以某种方式使用了Retrieve New。Prototype明确使用Retrieve New来确保始终返回对象的最新副本，而Builder则提供了一个极好的例子，展示了Retrieve New封装复杂创建逻辑的能力。抽象工厂和工厂方法在许多情况下也最终使用了Retrieve New 。
Many of the creational patterns from the original Design Patterns [21] use Retrieve New in some manner. Prototype uses Retrieve New explicitly to ensure that fresh copies of an object are always returned, and Builder provides an excellent example of Retrieve New’s ability to encapsulate complex creation logic. Abstract Factory and Factory Method also end up using Retrieve New in many cases.
检索共享
Retrieve Shared
对象管理
Object Management
意图
获取对共享对象的引用，但不拥有该对象的明确所有权。在面向对象编程中随处可见。
To obtain a reference to a shared object without holding explicit ownership of that object. Ubiquitous in object-oriented programming.
动机
对象是概念实体，任何特定实例都可能封装状态或提供可能同时引起许多其他对象兴趣的功能。例如，考虑一个打印机队列。许多应用程序都希望访问该队列，但每个应用程序都没有理由拥有自己的队列。事实上，由于不同的队列相互竞争，这肯定会导致最终的资源冲突。相反，一个队列可以同时在多个应用程序之间共享。来自不同对象的请求可以按照队列最适合的顺序进行处理。队列是一种共享资源，必须可供任何需要它的人使用。
Objects are conceptual entities, and any particular instance may encapsulate state or provide functionality that may be of interest to many other objects simultaneously. Consider a printer queue, for example. Many applications want to have access to the queue, but there is no reason for each application to have its own queue. In fact, this would certainly lead to eventual resource collisions as differing queues competed. Instead, a queue can be shared among many applications simultaneously. Requests from disparate objects can be handled in the order best suited by the queue. The queue is a shared resource, and must be available for whoever needs it.
Retrieve Shared同时使用Create Object和Retrieve ， Retrieve New也是如此，但不同之处在于它可以自由缓存或以其他方式将对新创建对象的引用移交给其他用途。不仅不能保证新创建的对象没有其他引用，而且明确假定它确实有其他引用。在持有引用的实体之间，所有权定义不明确。
Retrieve Shared uses both Create Object and Retrieve, as does Retrieve New, but differs in that it is free to cache or otherwise hand off references to the newly created object for other purposes. Not only is there no guarantee that the newly created object has no other references, it is explicitly assumed that it does have other references to it. Ownership is ill-defined among those entities holding references.
检索共享非常常见，许多语言都在某种程度上直接支持它。例如，C++ 具有该shared_ptr构造，它包含在最近采用的 2011 C++ 标准第 20.7.2.2 节中 [ 18 ]。
Retrieve Shared is so common that many languages support it directly on some level. C++, for example, has the shared_ptr construct, which is included in Section 20.7.2.2 of the recently adopted 2011 C++ standard [18].
其他语言和环境有惯用语和假定库支持对引用对象的准手动管理，例如 Objective-C 的autorelease功能。许多其他语言和环境使用垃圾收集在运行时自动检测并从系统中移除“死”对象。Java、C#、Python 和大多数脚本语言都是最后一种情况的例子。自动内存管理在编程语言中越来越流行和普遍，但它仍然远远不够来自 universal。所有开发人员都应该掌握手动内存管理的基础知识。
Other languages and environments have idioms and assumed library support for quasi-manual management of referenced objects, such as Objective-C’s autorelease feature. Many others use garbage collection to automatically detect and remove “dead” objects from the system during runtime. Java, C#, Python, and most scripting languages are examples of this last case. Automated memory management is becoming more popular and common among programming languages, but it is still far from universal. All developers should have the basics of manual memory management under their belt.
适用性
在以下情况下使用检索共享：
Use Retrieve Shared when:
• 远程对象提供本地计算所需的值对象，并且是：
• A remote object provides a value object that is required for local computation and is either:
– 由方法调用的返回值提供，或者
– provided by a method call’s return value or
– 由公开的字段对象提供。
– provided by an exposed field object.
• 该对象应该与其他对象共享，而不应被视为私有资源。
• The object should be shared with other objects and not considered a private resource.
结构
PINbox only version (identical to Retrieve New):
參與者
来源
Source
负责交出所检索对象的对象。它充当初始源，创建对象。
The object tasked with handing off the retrieved object. It serves as the initial source, creating the object.
下沉
Sink
target请求检索的对象并包括要赋予新值的项目的对象（或类）类型。
The object (or class) type that requests the retrieved object and includes the item, target, to be given a new value.
检索
Retrieved
要更新的值的类型和返回的值。
The type of the value to be updated and the value that is returned.
目标
target
更新到检索到的对象以供本地使用的字段。
The field that is updated to the retrieved object for local use.
已选定
selected
生成并返回新值的方法或字段。在这种情况下，返回的对象是对对象的引用，并且不保证该引用以任何方式与所有权相关联。
The method or field that produces and returns the new value. In this scenario, the returned object is a reference to an object, and the reference is not guaranteed to be associated with ownership in any way.
合作
此协作只有三个对象，它们仅扮演请求发起者、请求执行者和传递对象的角色。将此模式与“检索新内容”进行比较和对比。模式元素之间的差异非常小，但影响会迅速向外扩散。
This collaboration has only three objects and they merely play the parts of the request originator, the request fulfiller, and the passed object. Compare and contrast this pattern with Retrieve New. The differences between the elements of the patterns are extremely minor, but the effects can ripple outwards quickly.
结果
在垃圾收集语言或环境中，检索到的对象使用此模式获得引用计数，但不会限制新引用的生成。生成新引用很容易导致内存泄漏，因此请确保其他地方存在明确定义的所有权策略。如果需要以保证唯一性的方式检索对象，或者希望成为新分配的共享资源的主要所有者，请考虑使用“检索新” 。
In a garbage-collected language or environment, the retrieved object gains a reference count by using this pattern but does not restrict the generation of new references. Generating new references can lead to memory leaks fairly readily, so be certain a well-defined ownership policy exists elsewhere. Consider Retrieve New instead if an object needs to be retrieved with assured uniqueness or wishes to be the primary owner of a newly allocated shared resource.
执行
在 Java 中：
In Java:
class Retrieved {
2 };
4 class Source {
public Retrieved giveMeAValue() {
6 Retrieved ret = new Retrieved();
// 对 ret 执行某些操作，例如
8 // 缓存到其他地方
return ret;
10 };
};
12
class Sink {
14 Retrieved target;
Source srcobj;
16 public void operation() {
target = srcobj.giveMeAValue();
18 };
};
class Retrieved {
2 };
4 class Source {
public Retrieved giveMeAValue() {
6 Retrieved ret = new Retrieved();
// Do something with ret, such as
8 // cache elsewhere
return ret;
10 };
};
12
class Sink {
14 Retrieved target;
Source srcobj;
16 public void operation() {
target = srcobj.giveMeAValue();
18 };
};
相关模式
从结构图来看， Retrieve Shared和Retrieve New显然共享许多相同的概念管道。其区别性在于返回对象的所有权的唯一性。
Retrieve Shared and Retrieve New obviously share much of the same conceptual plumbing when looking at the structure diagrams. The distinguishing feature is the uniqueness of the ownership of the returned object.
单例可能是利用检索共享的最常见的设计模式。
Singleton is perhaps the most common design pattern that utilizes Retrieve Shared.
客体化者
Objectifier
本规范简要回顾了 Walter Zimmer 的Objectifier模式，并展示了它如何由单个较小模式组成。此模式的完整定义和原始讨论可在 [ 43 ] 中找到。如有必要，请参阅该文本以获取基础文档。
This specification briefly recaps Walter Zimmer’s Objectifier pattern and shows how it is a multiplicity of a single smaller pattern. The full definition and original discussion of this pattern can be found in [43]. Please refer to that text for the base document if necessary.
意图
Intent
根据原始规范，对象化器旨在“将其他类中的类似行为对象化，以便客户端可以独立于其他行为改变此类行为” [ 43，第 363 页]。请注意短语“其他类中的类似行为”。由此我们可能希望看到某种形式的重定向，但“在其他类中”限定表明我们可能正在处理子类化的情况。从实现方法中我们知道，只要抽象方法和具体方法之间的意图相似，我们就可以在子类中实现抽象方法。多个具体子类可能以略有不同的方式实现行为，但子类之间的整体意图保持相似。
Objectifier is, according to the original specification, intended to “objectify similar behaviour in additional classes, so that clients can vary such behaviour independently from other behaviour” [43, p. 363]. Note the phrase “similar behaviour in additional classes.” From this we might expect to see some form of Redirection, but the “in additional classes” qualification is a clue that we are probably dealing with a subclassing situation here. We know from Fulfill Method that we can implement an abstract method in a subclass as long as we have similar intent between the abstract and concrete methods. Multiple concrete subclasses may implement the behavior slightly differently, but the overall intent among the subclasses will remain similar.
动机
Motivation
我们在 Zimmer 最初的动机部分发现我们的分析是正确的。“设计中经常遇到的一个问题是抽象与其实现的分离，以及实现的互换”[ 43，第 363 页]。Zimmer 描述了一个具有抽象接口的超类，以及多个提供实现的子类。这告诉我们，我们有多个Fulfill 方法实例，但它们在抽象者角色中都共享一个共同的超类。可以使用堆叠的 PINbox 以图形方式显示这一点，正如我们在第 3 章第 3.2.4 节中看到的那样。
We find in Zimmer’s original Motivation section that our analysis was correct. “A frequent problem in design is the separation of an abstraction from its implementation, and the interchange of implementations” [43, p. 363]. Zimmer is describing a single superclass with an Abstract Interface, and multiple subclasses that provide implementations. This tells us that we have multiple Fulfill Method instances but that they all share a common super-class in the Abstractor role. This can be shown graphically using a stacked PINbox, as we saw back in Chapter 3, Section 3.2.4.
Zimmer 为上述子类实现集合添加了一个关键部分。他指定“为了将不同的行为客观化，[...] 具有可以在运行时互换的独立实现对象” [第 363 页]。他的意思是，开发人员应该创建对类型为该类的对象的引用Abstractor，然后根据需要交换子类的实际实例。现在，对该实例的所有调用都将以不可见的方式自动发送到正确的实现通过多态性。如何进行调用并不特别重要；重要的是某些客户端对象可以这样做。
Zimmer adds one critical piece to the above collection of subclass implementations. He specifies that “to objectify the varying behaviour, [...] have independent implementation objects that can be interchanged at runtime” [p. 363]. By this he means that the developer should create a reference to an object typed to be of the Abstractor class and then swap out actual instances of the subclass as needed. All calls made to that instance will now invisibly and automatically get sent to the correct implementation via polymorphism. How the call is made is not particularly relevant; the important thing is that some client object can do so.
适用性
Applicability
Zimmer 对Objectifier适用性的描述没有什么可补充的，因此我在此完整引用：
There is little to add to Zimmer’s description of the applicability of Objectifier, so I quote it here in full:
在以下情况下使用Objectifier模式
Use the Objectifier pattern when
• 行为应该与类分离，以获得可以交换、保存、更改、共享或调用的独立行为对象。
• Behaviour should be decoupled from classes to have independent behaviour objects which can be interchanged, saved, changed, shared or invoked.
• 需要运行时行为配置。
• Run-time configuration of behaviour is required.
• 有多个几乎相同的类，它们仅在一个或几个方法上有所不同。将不同的行为对象化到其他类中可以将以前的类统一到一个公共类中，然后可以通过引用新的附加类来配置该类
• There are several almost identical classes that differ only in one or a few methods. Objectifying the different behaviour in additional classes allows to unify the former classes in one common class, which can then be configured with a reference to the new, additional classes
• 有大量的条件代码来选择行为 [ 43，第 363 页]。
• There is a large amount of conditional code to select behaviour [43, p. 363].
在决定何时使用Objectifier时，有两点尤为突出。首先，需要运行时配置；其次，有大量的条件代码来选择行为。
Two items stand out here in deciding when to use Objectifier. First, that run time configuration is required, and second, that there is a large amount of conditional code to select behavior.
每当你遇到类似清单 6.1 的情况时，你都应该停下来考虑使用Objectifier。如果你发现多个这样的条件结构使用相同的触发器，那么这是一个强烈的信号，表明你可能能够将条件块重构为具有共同超类的类以提供接口。然后用请求方法的简单调用替换条件块，如清单 6.2所示。这看起来应该有点熟悉：它与第5 章抽象接口规范中提供的示例有关。
Any time you run across a situation that looks something like Listing 6.1, it should make you pause to consider using Objectifier. If you find multiple such conditional structures using the same triggers, then it is a strong sign that you may be able to refactor the conditional blocks into classes with a common superclass to provide an interface. The conditional blocks are then replaced with a simple invocation of the requested method, as in Listing 6.2. This should look a bit familiar: it is related to the example provided back in the specification for Abstract Interface in Chapter 5.
请注意， 的外部接口feedCritterInEnvirons()没有改变，但下面的代码现在更简洁了。添加新的动物类型只需要将其定义为 的子类，并在代码中选择动物种类的某个Animal位置进行更改。以前，添加新的动物类型需要在整个代码中分散向可能适用的每个位置添加条件块。feedCritterInEnvirons()
Note that the external interface to feedCritterInEnvirons() did not change, but the code underneath is now much cleaner. Adding a new animal type just requires defining it as a subclass of Animal and making a change at one point in the code, where the animal kind is being selected in feedCritterInEnvirons(). Previously, adding a new animal type would have required adding conditional blocks to each place that it might be applicable, scattered throughout the code.
Listing 6.1. Conditionals to select behavior.
1
typedef enum {鱼, 马, 猎豹} CritterKind;
3 CritterKind 小动物;
5 void eatFood(CritterKind 小动物, FoodItem f) {
if (critter == 鱼) {
7 swimToFood(critter, locationOfFood(f));
摄取(critter, f);
9 消化(critter, f);
} else if (critter == 马) {
11 checkHerdSafety(critter);
摄取(critter, f);
13 消化(critter, f);
} else if (critter == 猎豹) {
15 turboMode(critter, locationOfFood(f));
if (caughtPrey(f)) {
17 摄取(critter, f);
消化(critter, f);
19 }
} else {
21 摄取(critter, f);
消化(critter, f);
23 }
};
25
void ingest(CritterKind critter, FoodItem f) {
27 if (critter == FISH) {
...
29 } else if (critter == HORSE) {
...
31 } else if (critter == CHEETAH) {
...
33 } else {
...
35 }
};
37
typedef enum {OCEAN, PLAINS, SAVANNAH} Environment;
39
void feedCritterInEnvirons(Environment env) {
41 if (Environment == OCEAN) {
if (isHungry(FISH)) {
43 FoodItem f = findFood(FISH);
if (f) {
45 eatFood(FISH, f);
}
47 }
} else if (Environment == PLAINS) {
49 if (isHungry(HORSE)) {
FoodItem f = findFood(HORSE);
51 如果(f) {
eatFood(HORSE, f);
53 }
}
55 } else if (Environment == SAVANNAH) {
如果(isHungry(CHEETAH)) {
57 FoodItem f = findFood(CHEETAH);
如果(f) {
59 eatFood(CHEETAH, f);
}
61 }
}
63 };
1
typedef enum{FISH, HORSE, CHEETAH} CritterKind;
3 CritterKind critter;
5 void eatFood(CritterKind critter, FoodItem f) {
if (critter == FISH) {
7 swimToFood(critter, locationOfFood(f));
ingest(critter, f);
9 digest(critter, f);
} else if (critter == HORSE) {
11 checkHerdSafety(critter);
ingest(critter, f);
13 digest(critter, f);
} else if (critter == CHEETAH) {
15 turboMode(critter, locationOfFood(f));
if (caughtPrey(f)) {
17 ingest(critter, f);
digest(critter, f);
19 }
} else {
21 ingest(critter, f);
digest(critter, f);
23 }
};
25
void ingest(CritterKind critter, FoodItem f) {
27 if (critter == FISH) {
...
29 } else if (critter == HORSE) {
...
31 } else if (critter == CHEETAH) {
...
33 } else {
...
35 }
};
37
typedef enum {OCEAN, PLAINS, SAVANNAH} Environment;
39
void feedCritterInEnvirons(Environment env) {
41 if (Environment == OCEAN) {
if (isHungry(FISH)) {
43 FoodItem f = findFood(FISH);
if (f) {
45 eatFood(FISH, f);
}
47 }
} else if (Environment == PLAINS) {
49 if (isHungry(HORSE)) {
FoodItem f = findFood(HORSE);
51 if (f) {
eatFood(HORSE, f);
53 }
}
55 } else if (Environment == SAVANNAH) {
if (isHungry(CHEETAH)) {
57 FoodItem f = findFood(CHEETAH);
if (f) {
59 eatFood(CHEETAH, f);
}
61 }
}
63 };
Listing 6.2. Using Objectifier to select behavior.
1动物 类{
public:
3 void eatFood(FoodItem f) {
this ->ingest(f);
5 this ->digest(f);
};
7 void ingest(FoodItem f);
};
9
鱼类：公共动物 {
11 public:
void eatFood(FoodItem f) {
13 this ->swimToFood(locationOfFood(f));
this ->ingest(f);
15 this ->digest(f);
};
17 void ingest(FoodItem f);
};
19马
类：公共动物 {
21 public:
void eatFood(FoodItem f) {
23 checkHerdSafety(critter);
ingest(critter, f);
25 digest(critter, f);
};
27 void ingest(FoodItem f);
};
29
类Cheetah :公共动物 {
31 公共:
void eatFood(FoodItem f) {
33 turboMode(critter, locationOfFood(f));
if (caughtPrey(f)) {
35 ingest(critter, f);
digest(critter, f);
37 }
};
39 };
41 typedef enum {OCEAN, PLAINS, SAVANNAH} 环境;
43 void feedCritterInEnvirons(环境 env) {
动物* critter;
45 如果(环境 == OCEAN) {
critter = new Fish();
47 } else if (环境 == PLAINS) {
critter = new Horse();
49 } else if (环境 == SAVANNAH) {
critter = new Cheetah();
51 }
if (isHungry(critter)) {
53 FoodItem f = findFood(critter);
if (f) {
55 eatFood(critter, f);
}
57 }
};
1 class Animal {
public:
3 void eatFood(FoodItem f) {
this->ingest(f);
5 this->digest(f);
};
7 void ingest(FoodItem f);
};
9
class Fish : public Animal {
11 public:
void eatFood(FoodItem f) {
13 this->swimToFood(locationOfFood(f));
this->ingest(f);
15 this->digest(f);
};
17 void ingest(FoodItem f);
};
19
class Horse : public Animal {
21 public:
void eatFood(FoodItem f) {
23 checkHerdSafety(critter);
ingest(critter, f);
25 digest(critter, f);
};
27 void ingest(FoodItem f);
};
29
class Cheetah : public Animal {
31 public:
void eatFood(FoodItem f) {
33 turboMode(critter, locationOfFood(f));
if (caughtPrey(f)) {
35 ingest(critter, f);
digest(critter, f);
37 }
};
39 };
41 typedef enum {OCEAN, PLAINS, SAVANNAH} Environment;
43 void feedCritterInEnvirons(Environment env) {
Animal* critter;
45 if (Environment == OCEAN) {
critter = new Fish();
47 } else if (Environment == PLAINS) {
critter = new Horse();
49 } else if (Environment == SAVANNAH) {
critter = new Cheetah();
51 }
if (isHungry(critter)) {
53 FoodItem f = findFood(critter);
if (f) {
55 eatFood(critter, f);
}
57 }
};
结构
Structure
參與者
Participants
我们可以将以下内容添加到 Zimmer 提供的参与者中：
We can add the following to the participants that Zimmer provides:
手术
operation
正在抽象出来并由客户端请求的方法。
The method that is being abstracted out and requested by the client.
具体类
Concrete Class
提供实现的子类之一。
One of the subclasses providing an implementation.
合作
Collaborations
齐默的简短讨论没有什么可补充的，因此我在此将其全部引用。
There is little to add to Zimmer’s brief discussion, so I quote it here in its entirety.
• 客户端可以使用Objectifier委托部分行为。Objectifier在初始化期间接收完成任务所需的信息，或者客户端在调用 Objectifier时将信息作为参数传递。
• A client may use Objectifier to delegate parts of its behaviour. The Objectifier receives the information needed to fulfill its task during its initialization, or the client passes the information as a parameter when calling the Objectifier.
• 可以为客户端配置一个具体的对象器，以调整其行为，完成某项特定的任务 [ 43，第 364 页]。
• A client can be configured with a concrete Objectifier to adapt the behaviour to fulfil a certain task [43, p. 364].
结果
Consequences
总体而言，这种模式非常有用。它在大多数系统中提供了极大的灵活性和可扩展性，同时增强了代码主体的可理解性和清晰度。然而，如果意图行为没有明确说明，或者与类层次结构中不相关的行为交织在一起。请注意，您希望在Objectifier类系列中捆绑在一起的方法的意图是兼容且明确的。
Overall, this pattern is highly useful. It offers a great degree of flexibility and extensibility in most systems, while enhancing the comprehensibility and clarity of a body of code. It can, however, lead to confusion if the intent of behavior is not explicitly stated or is intertwined with unrelated behaviors within a class hierarchy. Be careful that the intents of methods you wish to bundle together in an Objectifier class family are compatible and clear.
执行
Implementation
任何支持Fulfill Method的语言都可以支持Objectifier。这种模式通常出现在大多数使用多态类结构的现代软件中。
Any language that supports Fulfill Method can support Objectifier. This pattern is typically found in most modern software that uses a class structure polymorphically.
相关模式
Related Patterns
Zimmer 列举了许多使用Objectifier的四人帮模式，包括Bridge、Builder、Command、Iterator、Observer、State和Strategy。除了这些之外，我们还可以通过下一个模式Object Recursion中的Objectifier来添加Chain of Responsibility、Composite和Decorator。
Zimmer enumerates many Gang of Four patterns that use Objectifier, including Bridge, Builder, Command, Iterator, Observer, State, and Strategy. To these we can add Chain of Responsibility, Composite, and Decorator, through the existence of Objectifier in the next pattern, Object Recursion.
对象递归
Object Recursion
本规范简要回顾了 Bobby Woolf 的对象递归，并说明它是两个较小模式的简单交集。此模式的完整定义和原始讨论可在 Woolf 的“对象递归模式” [ 41 ] 中找到。如有必要，请参阅该文本以获取基础文档。
This specification briefly recaps Bobby Woolf’s Object Recursion and demonstrates that it is a simple intersection of two smaller patterns. The full definition and original discussion of this pattern can be found in Woolf’s “The Object Recursion Pattern” [41]. Please refer to that text for the base document if necessary.
意图
Intent
Booby Woolf 对其意图的表述如下：“通过多态委托，将请求处理分散到结构中。对象递归透明地将请求反复分解为更易于处理的较小部分”[ 41 ，第41页]。然而，这并没有提到那些较小的部分，从我们对委托和重定向的讨论中，我们知道行为子任务之间的关系会显著影响系统的实现方式。阅读完整规范后，我们可以补充说，这些“较小的部分”在意图和行为上是密切相关的。
Booby Woolf states the intent as follows: “Distribute processing of a request over a structure by delegating polymorphically. Object Recursion transparently enables a request to be repeatedly broken into smaller parts that are easier to handle” [41, p. 41]. This says nothing about those smaller parts, however, and from our discussions of Delegation and Redirection, we know that the relationship between subtasks of a behavior can significantly affect how a system is implemented. After reading the entire specification, we can add that these “smaller parts” are closely related in intent and behavior.
动机
Motivation
在原始规范中，Woolf 为这种比较两个对象的模式提供了一个驱动示例，即让它们告诉各自的成员自己进行比较，以递归方式沿着对象树向下工作。这是一个很好的用例，说明了对象递归的一个必要组成部分：在每个级别执行的任务都是同一种工作。这是一个重定向的示例，因此我们希望在对象递归的任何分解中找到某种形式的重定向。
In the original specification, Woolf used a driving example for this pattern of comparing two objects by having them tell their respective members to compare themselves, recursively working down an object tree. This is a good use case and illustrates a necessary component of Object Recursion: that the task being performed at each level is the same kind of work. This is an example of redirection, so we expect to find some form of Redirection in any decomposition of Object Recursion.
在 Intent 部分中使用多态性一词向我们表明了两件事。首先，它告诉我们我们将看到至少一个Fulfill Method实例以及底层的Inheritance和Abstract Method EDP。其次，它告诉我们几乎肯定会有多个这样的实例，也就是说，每个潜在的子类目标都有一个实例。因此，我们期望看到Objectifier的一个实例。
The use of the word polymorphism in the Intent section indicates two things to us. First, it tells us that we’re going to see at least one instance of Fulfill Method as well as the underlying Inheritance and Abstract Method EDPs. Second, it tells us that there almost certainly will be more than one such instance, that is, one for each potential subclass target. We therefore expect to see an instance of Objectifier.
到目前为止，我们已经有了两个定义明确且简单的概念，可以用来帮助我们理解对象递归模式。我们只需要我们需要观察在继续分析这种模式时如何改进它们，以及它们如何结合在一起形成更复杂的行为和概念。
At this point, we have a duo of well-defined and simple concepts that we can use to help us understand the Object Recursion pattern. We just need to see how they might be refined as we continue analyzing this pattern and how they come together to form this more complex behavior and concept.
适用性
Applicability
Woolf 原始规范中的适用性部分指出，对象递归在以下情况下适用：
The Applicability section in Woolf’s original specification states that Object Recursion is appropriate when:
• 通过最终目的地未知的链接结构传递消息。
• passing a message through a linked structure where the ultimate destination is unknown.
• 向链接结构部分中的所有节点广播消息。
• broadcasting a message to all nodes in part of a linked structure.
• 在整个链接结构中分配行为的责任 [ 41 ，第44页]。
• distributing a behavior’s responsibility throughout a linked structure [41, p. 44].
值得注意的是，如果我们将适用性仅限于这些约束，那么无需使用 Woolf 提供的解决方案，问题也是可以解决的。具体来说，适用性中没有任何内容涉及分布式行为的相对意图，或对象可以相互信任以执行该行为的级别。如上所述，上述约束可以通过使用“重定向”或“委托”轻松由一连串不相关的对象处理，只需对标准进行一些创造性的解释即可。
It’s interesting to note that if we were to limit the applicability to only these constraints, the problem would be solvable without using the solution that Woolf provided. Specifically, nothing in the applicability addresses the relative intents of the behavior that is distributed or the level of trust that the objects can place in each other to perform that behavior. The above constraints, as stated, could just as easily be handled by a chain of unrelated objects through uses of Redirection, or, with a bit of creative interpretation of the criteria, Delegation.
从意图和动机部分的信息中，我们可以推断出对象递归的一些其他关键特征并增强本节。例如，我们知道多态性将占据突出地位，并且前面描述的“链接结构”可能有一些非常具体的要求，主要是链接结构由通过共同祖先相互关联的对象组成，如意图部分中提到的多态性所证明的那样。原始规范的后续部分证实了这一假设。
From the information in the Intent and Motivation sections, we can deduce some other key characteristics of Object Recursion and enhance this section. For instance, we know that polymorphism will figure prominently and that the “linked structure” described previously likely has some very specific requirements, primarily that the linked structure is composed of objects whose types are related to one another via a common ancestor, as evidenced by the polymorphism mentioned in the Intent section. This assumption is borne out by the later sections in the original specification.
这个共同祖先线索让我们期望看到我们的重定向的子类型关系形式。换句话说，我们可以指定，而不是通用重定向，我们将期望看到受信任的重定向。
This common ancestor clue leads us to expect to see the subtyping-relation form of our Redirection. In other words, we can specify that instead of a generalized Redirection, we will expect to see Trusted Redirection.
利用这些信息，我们可以按如下方式完善适用性：
Using this information, we can refine the applicability as follows:
• 存在诸如树之类的链接结构，并且结构中的元素具有共同的意图和目的，每个元素都有共同定义的行为。
• A linked structure such as a tree exists, and the elements in the structure have a shared intent and purpose, each with common defined behaviors.
• 需要通过此链接结构发送消息以调用其中一种常见行为；消息的意图在传递过程中需要保持基本不变。
• A message needs to be sent through this linked structure to invoke one of the common behaviors; the intent of the message needs to remain essentially unaltered in intent during passage.
• 在链接结构中，传递消息的最终目的地是未知的。
• The ultimate destination of the passed message is unknown within the linked structure.
• 消息可能只需要广播到链接结构的一部分中的所有节点。
• The message may need to be broadcast to all nodes in just a portion of the linked structure.
• 完成消息所请求的行为的责任分布在整个链接结构中。
• Responsibility for fulfilling the behavior requested by the message is distributed throughout the linked structure.
现在很明显，链接结构不包含随机元素，但结构中的每个节点在性质和意图上都是相关的。还明确指出，可以调用常见行为，行为可以通过结构中的对象分发，以及/或者行为可能适用于或不适用于整个结构。所有这些点对于塑造解决方案的最终面貌都很重要。
Now it is clear that the linked structure does not contain random elements, but each node in the structure is related in nature and intent. It is also clearly stated that common behaviors can be invoked, that behavior can be distributed through objects in the structure, and/or that the behavior may or may not be applicable to the entire structure. All of these points are important for shaping what the solution will eventually look like.
结构
Structure
我们可以在检查 Woolf 提供的原始 UML 图时使用这些推论，并使用 PINbox 对其进行注释，如图所示。正如我们推测的那样， Objectifier和Trusted Redirection实例确实存在。
We can use these deductions during an inspection of the original UML diagram provided by Woolf and annotate it with PINboxes, as shown. The Objectifier and Trusted Redirection instances do exist, as we surmised.
參與者
Participants
Woolf 的原始规范列出了四个参与者：发起者、处理程序、递归者和终止者。除此之外，为了结构 PIN 图和讨论的清晰度，我们可以再添加两个：handleRequest 和继任者。然后可以将参与者列为：
Woolf’s original specification lists four participants: Initiator, Handler, Recursor, and Terminator. To these, we can add two more for the structure PIN diagrams and for clarity in the discussion: handleRequest and successor. The participants then can be listed as:
发起者
Initiator
– 发起请求。
– initiates the request.
– 通常不是 的子类型Handler；makeRequest()是来自 的单独消息handleRequest()。
– usually not a subtype of Handler; makeRequest() is a separate message from handleRequest().
处理程序
Handler
– 定义一种可以处理发起者发出的请求的类型。
– defines a type that can handle requests that initiators make.
递归器
Recursor
– 定义successor链接。
– defines the successor link.
– 通过将请求委托给后继者来处理请求。
– handles a request by delegating it to its successors.
We can reduce this down to a much more readable PINbox-only version:
– 与某个请求相关的后继者可能因请求而异。
– successors relevant to a request can vary by request.
– 可以在委托请求之前或之后执行额外的行为。
– can perform extra behavior before or after delegating the request.
– 可能是不同请求的终止符。
– may be a terminator for a different request.
终结者
Terminator
– 通过完全实现请求而不是委托任何实现来完成请求。
– finishes the request by implementing it completely and not delegating any of its implementation.
– may be a recursor for a different request [41, p. 44].
处理请求
handleRequest
该方法在每个阶段处理所请求的行为并负责根据需要将其传递给后继者。
The method that handles the requested behavior at each stage and is responsible for passing it along to a successor as necessary.
接班人
successor
handleRequest从其实例接收消息的对象Recursor;是 类型Handler。
The object that receives the handleRequest message from an instance of Recursor; it is of type Handler.
合作
Collaborations
在原始 UML 序列图中添加 PINbox 后，我们可以看到对象器和受信任重定向在实现对象递归行为时的职责之间的准确划分。它们各自占据序列图的一半，并在aRecursor对象中相遇。
Adding PINboxes to the original UML sequence diagram lets us see the exact demarcation between the responsibilities of Objectifier and Trusted Redirection in implementing the behavior of Object Recursion. Each is taking half of the sequence diagram, and they meet in the aRecursor object.
结果
Consequences
Woolf 在“后果”中列出的一个缺点是：“编程复杂性。递归，无论是面向过程的还是面向对象的，都是一个难以掌握的概念。过度使用会使系统更难理解和维护”[ 41 ，第46页]。虽然确实如此，但可以通过将这种递归风格的两个基本元素划分为通过Objectifier实现的多态性结构和通过Trusted Redirection对该多态系统的递归式调用来澄清情况。现在明确展示了这种分布式递归技术的两个方面，有助于理解。
The one disadvantage that Woolf lists under Consequences is: “Programming complexity. Recursion, procedural or object-oriented, is a difficult concept to grasp. Overuse can make a system more difficult to understand and maintain” [41, p. 46]. While true, the situation can be clarified by dividing the two essential elements of this style of recursion into the structure of the polymorphism via Objectifier and the recursive-style call into that polymorphic system via Trusted Redirection. Now the two aspects of this distributed recursion technique are explicitly shown, aiding comprehension.
执行
Implementation
原始对象递归规范在其实现部分中指出，“Initiator.makeRequest() 消息不能是多态的使用 Recursor.handleRequest() 消息” [ 41 ，第47页]。前面的讨论明确说明了Objectifier和Trusted Redirection实例，清楚地表明了这种区别。
The original Object Recursion specification states in its Implementation section that “The Initiator.makeRequest() message must not be polymorphic with the Recursor.handleRequest() message” [41, p. 47]. The earlier discussion, where the Objectifier and Trusted Redirection instances are made explicit, shows this distinction clearly.
相关模式
Related Patterns
对象递归通常用于具有家族类簇结构的设计模式，包括责任链、装饰器、复合和解释器。抽象工厂可以使用对象递归来构建多层对象，每个子组件根据需要通过工厂接口请求适当的回调。
Object Recursion is commonly used in design patterns with a family class cluster structure, including Chain of Responsibility, Decorator, Composite, and Interpreter. Abstract Factory can use Object Recursion to enable the building of multilayered objects, with each subcomponent requesting an appropriate callback through the factory interface as needed.
7.四人帮图案构图
7. Gang of Four Pattern Compositions
至此，您已经了解了完整的方法调用 EDP 目录以及通过组合这些模式形成的几个示例。在本章中，我们将该过程提升到一个新的水平，展示如何组合第 2 章和第 6 章中的模式以形成开发人员最熟悉的几个四人帮 (GoF) 设计模式。
At this point, you have seen the full method-call EDP catalog and several examples of patterns that are formed by combining them. In this chapter, we take that process to the next level, showing how the patterns in Chapters 2 and 6 can be combined to form a few of the Gang of Four (GoF) design patterns that developers are most familiar with.
本章讨论了六种 GoF 模式，其中两种来自《设计模式》中定义的创建型、结构型和行为型系列。这些额外的模式为在这一级别上使用其他设计模式奠定了基础。为什么只有六种？虽然每种设计模式都可以用 EDP 来描述，但并非所有设计模式都能仅使用本书介绍的EDP 来充分描述。请记住，本书仅介绍基于方法调用的 EDP，而方法调用只是面向对象编程实体之间四种依赖关系之一。请重新参阅第 2 章第2.2.2 节的表 2.3以了解其余内容。之所以选择这六个，是因为它们与该组中的其他模式有着很好的关联，这种关联乍一看并不明显，但一旦揭示了底层概念结构，就会变得明显。
Six GoF patterns are discussed in this chapter, two each from the Creational, Structural and Behavioral families as defined in Design Patterns. These additional patterns provide a foundation for working with other design patterns on this level. Why only six? Although every design pattern can be described in terms of EDPs, not all design patterns can be adequately described using only the EDPs presented in this book. Remember that this book only covers EDPs based on method calls, and method calls are only one of the four kinds of reliance relationships among entities of object-oriented programming. Refer back to Chapter 2, Table 2.3, in Section 2.2.2 for a refresh on the remainder. These six were selected because they are nicely related to others in the group in ways that are not obvious at first glance but become apparent once the underlying conceptual structure is revealed.
本章中的讨论不遵循前两章使用的模式规范模板。这些设计模式在众多来源中都有很好的记录，这里没有必要重复结构。相反，讨论以更随意的方式概述了每个设计模式和组成。当然，PIN 和 UML 图用于说明相关关系和连接。
The discussions in this chapter do not follow the pattern specification template that the previous two chapters used. These design patterns are well documented in numerous sources, and there is no need to duplicate the structure here. Instead, the discussions outline each design pattern and the composition in a more casual way. PIN and UML diagrams, of course, are used to illustrate relevant relationships and connections.
7.1. 创建模式
7.1. Creational Patterns
创建型模式都与对象创建和管理有关。正如您所预料的，创建对象在其定义中占有重要地位，检索相关模式也是如此。这两个 EDP 如何相互连接以及与其他 EDP 连接，创建型模式之间形成了广泛的行为。我们分解和比较的两个是抽象工厂和工厂方法。它们提供了一个有趣的提醒，即组合上的微小差异如何导致实现和意图的巨大变化。
The creational patterns are all concerned with object creation and management. As you might expect, Create Object features heavily in their definitions, as do the Retrieve-related patterns. How these two EDPs are connected to each other and to the other EDPs creates a wide array behaviors that distinguish the Creational patterns from each other. The two we decompose and compare are Abstract Factory and Factory Method. They provide an interesting reminder of how small differences in composition can lead to large changes in implementation and intent.
7.1.1. 抽象工厂
7.1.1. Abstract Factory
第一个创建模式抽象工厂是一个完美的例子，它展示了简单的概念如何以复杂的方式关联起来，从而产生一个有趣的概念。只需要三个基本模式就可以捕捉抽象工厂的大部分语义：实现方法、检索和创建对象。它们的连接方式决定了丰富的行为。
The first creational pattern, Abstract Factory, is a perfect example of how simple concepts can relate in complex ways to produce an interesting concept. Only three basic patterns are needed to capture most of the semantics of Abstract Factory: Fulfill Method, Retrieve, and Create Object. The manner in which they connect establishes the rich behavior.
抽象工厂的意图是“提供一个接口，用于创建相关或依赖对象系列，而无需指定它们的具体类”[ 21 ，第87页]。从这句话中，我们可以推断，我们很可能会看到使用继承来生成该接口，并使用实现方法来提供相关方法的实现。
The intent for Abstract Factory states that it “provides an interface for creating families of related or dependent objects without specifying their concrete classes” [21, p. 87]. From this statement, we can deduce that we are likely to see uses of Inheritance to produce that interface and of Fulfill Method to provide the implementation of the relevant methods.
图 7.1显示了扩展 PINbox 中的抽象工厂的结构图。PINbox 边界中的每个角色都只连接到一个指向抽象工厂的元素互连：createProductA()在 中实现的方法ConcreteFactory2用于生成ProductA2具体类的实例，该实例是从该类中派生出来的AbstractProductA。其他三个产品类也存在类似的连接。
Figure 7.1 shows the structure diagram for Abstract Factory wrapped within an expanded PINbox. Each role in the PINbox border is connected to just one interconnection of elements that lead to Abstract Factory: the createProductA() method as implemented in ConcreteFactory2 is used to produce an instance of the ProductA2 concrete class, which is subclassed from the AbstractProductA class. Similar connections exist for the other three product classes.
图 7.1.扩展的 PINbox 中包含抽象工厂。
Figure 7.1. Abstract Factory subsumed within the expanded PINbox.
这是一个相当复杂的图表，很难准确地看出其中的关系。图 7.2更清楚地显示了单个互连，其中我们删除了结构图中的元素，这些元素不直接涉及这个单一抽象工厂实例。图 7.3通过稍微清理 PINbox 的内部结构进一步简化了它，最终将我们引向图 7.4 ，它说明了抽象工厂核心的简单概念结构。
This is a rather complex diagram, and it is difficult to see exactly what the relationships at work are. The single interconnection is shown more clearly in Figure 7.2, in which we have removed the elements of the structure diagram that are not directly involved in this single Abstract Factory instance. Figure 7.3 simplifies it further by cleaning up the internals of the PINbox a bit, leading us finally to Figure 7.4, which illustrates the simple conceptual structure at the core of Abstract Factory.
图 7.2. 将图表简化为仅有一个抽象工厂实例。
Figure 7.2. Reducing the diagram to just one instance of Abstract Factory.
图 7.3. 简化图 7.2。
Figure 7.3. Simplifying Figure 7.2.
图 7.4.抽象工厂仅作为 PIN。
Figure 7.4. Abstract Factory as PIN only.
此时，我们可以看到，这里存在一个相当简单的连接。Retrieve正在从Create Object返回一个实例，并通过调用通过Fulfill Method在子类中实现的抽象方法来实现。虽然看起来我们可以简单地将 Retrieve 和 Create Object 的实例合并为Retrieve New，但在某些情况下，抽象工厂可能会使用Flyweight等模式来节省空间地支持它所返回的对象，从而改为实现Retrieve Shared 。还要记住，这两种形式的Retrieve具有相同的 PIN 图，因此我们不能仅凭这一点来区分它们。由于我们无法提前确定这种情况，而应该留给设计者或实现者，因此我们此时保持两个 EDP 可见，以便清晰和灵活。
At this point, we can see that a fairly simple connection is at work here. Retrieve is returning an instance from Create Object and is doing so through a call to an abstract method that is being implemented in a subclass via Fulfill Method. Although it may look like we could simply collapse the instances of Retrieve and Create Object into a Retrieve New, in some situations, Abstract Factory may use a pattern such as Flyweight for space-efficient backing of the objects it is returning, resulting in Retrieve Shared being implemented instead. Remember also that these two forms of Retrieve have identical PIN diagrams, so we can’t distinguish between them on this basis alone. Because we can’t determine this situation ahead of time but should leave up to the designer or implementor, we keep the two EDPs visible at this point for clarity and flexibility.
7.1.2. 工厂方法
7.1.2. Factory Method
此模式将实际对象的创建推迟到子类，同时提供用于创建对象并在内部使用它的固定接口和基方法。在开始分析之前，我们认识到由于这是一种创建型模式，因此我们期望看到Create Object和Retrieve同时使用。读取Factory Method的 Participants 部分中的项目为我们提供了继续分解的必要线索。例如，Creator 的描述指出它“声明了工厂方法”[ 21 ，第108页]，而 ConcreteCreator“重写了工厂方法”。这是一个明显的Fulfill Method示例。它还指出 Creator“可以调用工厂方法”。这里有一些潜台词，但我们只要稍加思考就能弄清楚这对我们的分析意味着什么。如果 Creator 正在调用它声明的工厂方法，那么它只能通过两种方式之一来进行。要么工厂方法调用自身，要么工厂方法被另一个方法调用。前者是递归的一个例子，并且非常重要，如果它是工厂方法协作的一部分，几乎肯定会被提及。这使我们相信工厂方法被另一个方法调用，作为聚合的一个实例。检查工厂方法的结构 UML 图（图 7.5）证实意图是后一种情况。
This pattern defers creation of an actual object to a subclass while providing a firm interface and a base method for creating the object and using it internally. To begin our analysis, we recognize that because this is a Creational pattern, we expect to see both Create Object and Retrieve in use. Reading off the items from Factory Method’s Participants section gives us the necessary cues to continue the decomposition. For instance, the description for Creator states that it “declares the factory method” [21, p. 108], whereas ConcreteCreator “override[s] the factory method.” This is a clear example of a Fulfill Method. It is also indicated that Creator “may call the factory method.” There’s a bit of subtext here, but we can figure out what this means for our analysis with a bit of thought. If Creator is calling the factory method that it declares, it can only do so in one of two ways. Either the factory method is calling itself or the factory method is being called from another method. The former is an example of Recursion and is important enough that it almost certainly would be mentioned by name if it were part of the collaborations of Factory Method. This leads us to believe that the factory method is being called from another method, as an instance of Conglomeration. Inspecting the structure UML diagram for Factory Method (Figure 7.5) confirms that the intent is the latter case.
图 7.5.扩展的 PINbox 中包含的工厂方法。
Figure 7.5. Factory Method subsumed within the expanded PINbox.
工厂方法将 EDP创建对象、检索、实现方法和聚合编织在一个类型层次结构中，如图 7.5所示。通过将其简化为图 7.6中的 PIN 专用图，可以更清晰地了解该结构。
Factory Method weaves the EDPs Create Object, Retrieve, Fulfill Method, and Conglomeration together within one type hierarchy, as shown in Figure 7.5. This structure can be made clearer by reducing it to the PIN-only diagram in Figure 7.6.
图 7.6.工厂方法仅作为 PIN。
Figure 7.6. Factory Method as PIN only.
工厂方法是一个有趣的例子，因为它是一种 GoF 模式，但在其定义中明确使用了另一种 GoF 模式。GoF 在对工厂方法的讨论中提到，“工厂方法通常在模板方法中调用”[ 21 ，第116页]。现在我们清楚地了解了它们是如何相互作用的。本章末尾的图 7.19和随附的模板方法讨论解释了如何用等效的模板方法替换Fulfill Method和Conglomeration的实例。工厂方法是模板方法的扩展，它添加了对象的创建和检索。
Factory Method is an interesting example in that it is a GoF pattern that explicitly uses another GoF pattern in its definition. In their discussion of Factory Method, GoF mentions that “Factory methods are usually called within Template methods” [21, p. 116]. Now we have a clear understanding of how they interact. Figure 7.19 at the end of this chapter and the accompanying discussion of Template Method explain how the instances of Fulfill Method and Conglomeration could be replaced with an equivalent Template Method. A Factory Method is an extension of a Template Method that adds the creation and retrieval of an object.
7.2. 结构模式
7.2. Structural Patterns
从结构组中选出的两种模式是Decorator和Proxy。尽管它们的意图和表面结构截然不同，但它们却有着一些令人惊讶的共同点。
The two patterns selected from the Structural group are Decorator and Proxy. Despite their quite different intents and apparent structures, they share some surprising commonalities.
7.2.1. 装饰器
7.2.1. Decorator
装饰器是第 4 章第4.2 节中的驱动模式，您在那里看到了有关如何分析此模式的详细讨论。现在我们可以在其配套设计模式中为其提供适当的上下文。图 7.7将第 4.2 节中的图 4.17封装在 PINbox 中。
Decorator was the driving pattern in Chapter 4, Section 4.2, and there you saw a detailed discussion of how to analyze this pattern. Now we can give it a proper context within its companion design patterns. Figure 7.7 encapsulates Figure 4.17 from Section 4.2 in a PINbox.
图 7.7.装饰器被扩展的 PINbox 所涵盖。
Figure 7.7. Decorator subsumed with the expanded PINbox.
通过将图 7.7简化为图 4.18中的 PIN 图，类似地包装，我们可以生成一个简单的图表，如图7.8所示。现在底层概念已经清楚地表述出来，它们之间的联系也很明确。
By reducing Figure 7.7 to just the PIN diagram from Figure 4.18, similarly wrapped, we can produce a simple diagram, as in Figure 7.8. Now the underlying concepts are clearly stated, and their connections to each other are explicit.
图 7.8.装饰器仅作为 PIN。
Figure 7.8. Decorator as PIN only.
这个简单明了的图向我们展示了由对象递归和扩展方法组成的装饰器的内部组成。然而，为了便于讨论，进一步解构是有用的。我们可以将图 4.22中的信息以扁平形式呈现，如图7.9所示。现在，我们剥离了对象递归和对象化器PINbox，并直接暴露了实现方法和受信任重定向实例。
This simple and straightforward figure shows us the internal composition of Decorator from Object Recursion and Extend Method. For the purposes of this discussion, however, it is useful to continue the deconstruction further. We can present the information from Figure 4.22 in a flattened form, as in Figure 7.9. We’ve now peeled off the Object Recursion and Objectifier PINboxes and directly exposed the Fulfill Method and Trusted Redirection instances.
图 7.9。装饰器扩展为三层深度并变平。
Figure 7.9. Decorator expanded three levels deep and flattened.
7.2.2. 代理
7.2.2. Proxy
代理限制对对象或对象结构的访问，向客户端隐藏详细信息。这是一个非常简单的模式，我们只需查看结构图即可进行分析，如图7.10所示。我们可以很容易地看到有一个Fulfill MethodProxy::request()的实例，并且对同一方法的调用是Deputized RedirectionRealSubject的实例。
Proxy limits access to an object or object structure, hiding details from the clients. It is a simple enough pattern that we can perform our analysis by simply looking at the structure diagram, as shown in Figure 7.10. We can easily see that there is an instance of Fulfill Method, and the call from Proxy::request() to the same method in RealSubject is an instance of Deputized Redirection.
图 7.10.代理被扩展的 PINbox 所涵盖。
Figure 7.10. Proxy subsumed with the expanded PINbox.
使用代理重定向是此模式的重要特征。它表明我们将任务重定向到的对象属于高度可信的类型。为了清晰起见，我们可以将此图简化为仅显示 PIN（如图7.11所示） ，然后稍微重新组织一下此图（如图7.12所示） 。
The use of Deputized Redirection is the important characteristic of this pattern. It indicates that the objects to which we are redirecting a task are of highly trusted types. We can reduce this diagram to just the PIN for clarity, in Figure 7.11, and then reorganize the diagram a bit, as in Figure 7.12.
图 7.11.代理仅作为 PIN。
Figure 7.11. Proxy as PIN only.
图 7.12。重新组织代理PIN 以更好地匹配装饰器。
Figure 7.12. Proxy PIN reorganized to better match Decorator.
进行改组的原因是，现在Proxy的定义与我们在图 7.9中从对象递归中扩展出来的Decorator部分更加接近。如果仔细查看这两者，我们就会发现Proxy本质上是对象递归的一种高度可信的变体，但有一个显著的区别。当接收对象的实例放在其中一个子类中时，对象递归的深度递归特性就会被破坏。取而代之的是单一链接，即从代理到最终对象的一次性重定向。此外，我们可以看到Proxy缺少Decorator所具有的Extend Method实例。这表明，虽然Decorator将增加行为作为主要考虑因素，但Proxy却没有。这两点都得到了 GoF 的证实：“与 Decorator 不同，Proxy 模式不关心动态地附加或分离属性，而且它不是为递归组合而设计的”[ 21 ，第200页]。这些差异现在在两种模式的 PIN 图和组成中清楚地显示了出来。
The reason for the shuffling is that now the definition for Proxy more closely matches the section of Decorator that we expanded out of Object Recursion in Figure 7.9. If we look at those two closely, we can see that Proxy is essentially a highly trusted variant of Object Recursion, with one significant difference. When the instance of the recipient object is placed within one of the subclasses, the deeply recursive nature of Object Recursion is disrupted. Instead, we have a single linkage, a one-time redirection from the proxy to the final object. Additionally, we can see that Proxy lacks the instance of Extend Method that Decorator has. This indicates that while Decorator has the addition of behavior as a prime consideration, Proxy does not. These two points are verified by GoF: “Unlike Decorator, the Proxy pattern is not concerned with attaching or detaching properties dynamically, and it’s not designed for recursive composition” [21, p. 200]. These differences are now clearly shown in the PIN diagrams and compositions of these two patterns.
7.3. 行为模式
7.3. Behavioral Patterns
GoF 的行为模式提出了一个有趣的挑战，因为本文详述的四种基本依赖中只有方法调用依赖，而我们仅依赖方法调用 EDP。然而，如第2.2.2 节中所述，由我们的三个轴定义的设计空间提供了足够的上下文信息，因此可以对其中许多模式给出令人惊讶的稳健定义。特别是，责任链和模板方法值得研究。
The Behavioral patterns from GoF provide an interesting challenge given that the only one of the four kinds of basic reliances fleshed out in this text is the method-call reliance, and we’re relying solely on the method call EDPs. The design space defined by our three axes, as in Section 2.2.2, however, provides enough contextual information that many of these patterns can be given surprisingly robust definitions. In particular, Chain of Responsibility and Template Method are worth investigating.
7.3.1. 责任链
7.3.1. Chain of Responsibility
责任链模式与代理和装饰器有很多共同之处。我们像往常一样首先展示包裹在 PINbox 中的结构图，如图7.13所示。整体结构相当简单，仅由两个 EDP 组成。有多个扩展方法实例，但此图足够简单，我们可以放弃中间步骤并直接跳到仅 PIN 的图，如图7.14所示。
Chain of Responsibility is a pattern that has much in common with Proxy and Decorator. We start as usual by showing the structure diagram wrapped in a PINbox, as in Figure 7.13. The overall construct is fairly straightforward, being composed of only two EDPs. There are multiple instances of Extend Method, but this diagram is simple enough that we can forego the intermediate step and jump directly to the PIN-only diagram, as in Figure 7.14.
图 7.13.扩展的 PINbox 中包含的责任链。
Figure 7.13. Chain of Responsibility subsumed within the expanded PINbox.
图 7.14.责任链仅为 PIN。
Figure 7.14. Chain of Responsibility as PIN only.
实际上，如果你检查 EDP 衍生的责任链定义与装饰器定义，就会发现显著的区别在于责任链的重定向递归取代了装饰器的对象递归。如果使用责任链的变体之一，使得s 明确定义后继者 [ 21，第225页]，那么对象递归也会在这里找到。那么，唯一的区别就是模式中对象递归发生的位置。这是在极小的ConcreteHandler概念上的变化导致了截然不同的结果。责任链和装饰器在概念上非常接近，但最终却大不相同，以至于它们的创建者甚至没有将它们归为同一类别。
In fact, if you inspect the EDP-derived definition of Chain of Responsibility against Decorator, it becomes apparent that the salient difference is that the Redirected Recursion of Chain of Responsibility replaces the Object Recursion of Decorator. If one of the variants on Chain of Responsibility is used such that the ConcreteHandlers define the successor explicitly [21, p. 225], then Object Recursion will be found here as well. The only difference, then, is where in the pattern Object Recursion occurs. It is an interesting exercise in extremely small conceptual changes resulting in greatly different outcomes. Chain of Responsibility and Decorator are extremely close in their conceptual makeup, yet so different in the end that they were not even placed the same category by their creators.
7.3.2. 模板方法
7.3.2. Template Method
最后，模板方法是另一个很好的例子，它以一种非常特殊的方式结合两个简单的概念，以产生更强大的结构。
Finally, Template Method is another excellent example of combining two simple concepts in a very specific way to produce a much more powerful construct.
图 7.15显示了封装在扩展的 PINbox 中的原始结构图。您可以看到Conglomeration和Fulfill Method的多个实例，因此为了清晰起见，我们将其简化为一个简单的组合，如图7.16所示。请注意， Conglomeration的 operation2和Fulfill Method的操作指向同一种方法：primitiveOperation1()的抽象AbstractClass。
Figure 7.15 shows the original structure diagram encapsulated within an expanded PINbox. You can see multiple instances of both Conglomeration and Fulfill Method, so let’s reduce it down to a simple combination for clarity, as in Figure 7.16. Notice that operation2 of Conglomeration and operation of Fulfill Method point to the same method: the abstract primitiveOperation1() of AbstractClass.
图 7.15.扩展的 PINbox 中包含的模板方法。
Figure 7.15. Template Method subsumed within the expanded PINbox.
图 7.16。模板方法简化为单个实例。
Figure 7.16. Template Method reduced to a single instance.
现在可以轻松提取生成的 PIN 图。图 7.17清楚地显示了Conglomeration实例正在调用一个抽象方法，该方法将由子类通过 Fulfill Method 处理。Conglomeration的调用方法是，它调用的方法是同一个类中的抽象方法。这必然意味着这个类是抽象的，并且是Template Method和Fulfill Method的抽象类，这确保子类将提供适当的功能。因此，这个子类满足了Fulfill Method的具体类角色和Template Method的相同角色。但是，要求Conglomeration的目标方法是Fulfill Method实例中的抽象方法，会强制子类templateMethodprimitiveOperation以某种方式处理细节，以便可以动态调整算法各部分的实现。
Now the resultant PIN diagram can be extracted easily. Figure 7.17 clearly shows that the Conglomeration instance is calling into an abstract method that will be handled by a subclass through Fulfill Method. The calling method of Conglomeration is the templateMethod, and the method it is calling is the abstract primitiveOperation method in the same class. Necessarily, this means that this class is abstract, and the Abstract Class of both the Template Method and the Fulfill Method, which ensures that a subclass will provide the proper functionality. This subclass therefore fulfills the Concrete Class role of Fulfill Method and the same role of Template Method. Requiring the target method of a Conglomeration to be the abstract method in an instance of Fulfill Method, however, forces a subclass to handle the details in a way such that the implementation of portions of the algorithm can be adjusted dynamically.
图 7.17.模板方法仅作为 PIN。
Figure 7.17. Template Method as PIN only.
图 7.18显示了相同的 PIN 图信息，重新组织后再次显示了与我们之前讨论的三种模式的一致性。现在您可以看到，在Proxy使用Fulfill Method和Deputized Redirection的实例的地方，Template Method将后者替换为Conglomeration的一个实例。这消除了Proxy的受信任对象方面，而是将功能拉入单个对象中。但是，值得注意的是，在Proxy模式中，主要关系是 Proxy 元素中实现的抽象方法正在启动Deputized Redirection关系。相反，在Template Method中，主要特征是 Concrete Class 元素中实现的抽象方法是Conglomeration实例的目标。从Deputized Redirection到Conglomeration的转变以及方法调用关系的发起者和目标的交换都不是复杂或不寻常的，但它们一起导致了截然不同的结果。
Figure 7.18 shows the same PIN diagram information, reorganized to again show congruences with our previous three pattern discussions. Now you can see that where Proxy uses instances of Fulfill Method and Deputized Redirection, Template Method replaces the latter instance with one of Conglomeration. This eliminates the trusted objects aspect of Proxy and instead pulls the functionality within the single object. However, it is worth noting that in the Proxy pattern, the primary relationship is that the abstract methods fulfilled in the Proxy element are initiating the Deputized Redirection relationship. In contrast, in Template Method the primary feature is that the abstract methods fulfilled in the Concrete Class element are the targets of the Conglomeration instances. Neither the shift from Deputized Redirection to Conglomeration nor swapping of the initiator and target of the method-call relationship is complex or unusual, but together they lead to profoundly different results.
图 7.18。重新组织模板方法PIN 以更好地匹配装饰器。
Figure 7.18. Template Method PIN reorganized to better match Decorator.
在讨论工厂方法时，您了解到模板方法是一种可能的实现特性，现在您可以看到它是如何发挥作用的。图 7.19显示了重新定义的工厂方法，其中模板方法取代了先前的单个实现方法和聚合实例。现在，这两种模式之间的关系定义清晰，显而易见。工厂方法使用模板方法来灵活地处理基础结构，在其中嵌入了检索和创建对象的组合使用，就像在抽象工厂中所做的那样。
In the discussion of Factory Method, you learned that Template Method was a possible implementation feature, and now you can see how this fits in. Figure 7.19 shows Factory Method redefined with Template Method replacing the previously individual instances of Fulfill Method and Conglomeration. Now the relationship between these two patterns is clearly defined and obvious. Factory Method uses a Template Method to flexibly handle the infrastructure within which it embeds a combined use of Retrieve and Create Object, much as is done in Abstract Factory.
图 7.19。使用模板方法重新定义工厂方法。
Figure 7.19. Factory Method redefined using Template Method.
7.4. 结论
7.4. Conclusion
在本章中，我们汇集了前几章中的概念和技术，综合了您在实践中会遇到的最常见设计模式的分析。讨论结合了 EDP 和中间模式有意义的方法来揭示程序设计高级概念的底层概念和结构。此外，您还学习了如何使用 PIN 以多种变化形式说明这些模式，具体取决于您希望表达的细节或上下文级别。您看到了设计模式规范中的线索和提示如何指导您利用您对其他模式的现有知识来帮助您更好地了解新模式的功能。您还了解了即使在这个更高级别上，设计模式也可以如何相互使用，例如隐藏在工厂方法中的模板方法实例。
In this chapter, we drew together the concepts and techniques from the prior chapters, synthesizing analyses of the most common design patterns you will encounter in your practice. The discussions combined the EDPs and intermediate patterns in meaningful ways to reveal the underlying concepts and structure of high-level concepts of program design. Furthermore, you learned how to illustrate these patterns with PIN in a number of variations depending on what level of detail or context you wish to express. You saw how clues and cues in design pattern specifications can point you to leveraging your existing knowledge of other patterns to help you form a better picture of what a new pattern does. You also gained an understanding of how design patterns, even at this higher level, can be used with and inside each other, such as the instance of Template Method buried inside Factory Method.
至此，我们结束了对 EDP、PIN 的介绍，我希望这能为您带来一种新颖有趣的方式，让您了解我们行业中丰富的设计模式。现在，您已经掌握了一套工具和技术，您可以使用它们来进一步提高自己在软件设计方面的知识和技能，并通过帮助记录我们所拥有的内容来加深我们对设计模式的集体理解。
This concludes your introduction to EDP, PIN, and what is, I hope, a new and interesting way for you to look at the rich tapestry of design patterns available in our industry. You’re now armed with a set of tools and techniques that you can use to further your own knowledge and skill in software design and to further our collective understanding of design patterns by helping document what we have.
欢迎加入社区。我期待看到我们接下来的行动。
Welcome to the community. I look forward to seeing what we do next.
A.ρ-微积分
A. ρ-Calculus
本附录旨在简要介绍基本设计模式、rho 演算或 ρ 演算背后的形式化方法。对于那些热衷于理论或研究并对更完整的论文感兴趣的人，以下出版物提供了这些主题的正式论述。我希望您会发现它们是通往更丰富的编程可能性世界的合适切入点。阅读本讨论时，偶尔回顾第 2 章可能会很有用。
This appendix is designed to be a brief introduction to the formalisms behind the Elemental Design Patterns, the rho-calculus, or ρ-calculus. For those of you with an inclination for theory or research and who are interested in a more complete treatise, the following publications provide formal treatments of the topics. I hope you find them a suitable entry point to a richer world of programming possibilities. You may find it useful to refer back to Chapter 2 occasionally while reading this discussion.
• 马丁·阿巴迪和卢卡·卡德利。对象理论。施普林格出版社，纽约，1996 年。
• Martín Abadi and Luca Cardelli. A Theory of Objects. Springer-Verlag, New York, 1996.
• Jason McC. Smith。SPQR ：从源代码自动检测设计模式的正式基础和实际支持。博士论文，北卡罗来纳大学教堂山分校，2005 年 12 月。
• Jason McC. Smith. SPQR: Formal Foundations and Practical Support for the Automated Detection of Design Patterns From Source Code. PhD thesis, University of North Carolina at Chapel Hill, Dec 2005.
在2.2.2 节中，我们了解到，所有面向对象编程都可以使用四个要素来建模：对象、方法、字段和类型。方法、字段和类型是相当明显的组成部分，因为它们分别代表编程的三个基本元素：功能、数据存储和数据描述。无论语言范式如何，没有这三个元素就无法进行计算。我们为什么添加对象当时被抛在一边。
In Section 2.2.2, you were introduced to the idea that all of object-oriented programming can be modeled using just four things: objects, methods, fields, and types. Methods, fields, and types are rather obvious pieces to include because they represent, respectively, the three basic elements of programming: functionality, data storage, and data description. No computation can occur without all three of those elements, regardless of language paradigm. Why we added objects was sort of waved aside at the time.
尽管它可能仍显得自指，但添加对象的正式原因是它们构成了面向对象编程的核心和焦点。没有对象，就没有面向对象编程。不满意吗？请参考 Abadi 和 Cardelli 的 ς-演算（sigma-演算）。[ 1 ] ς-演算是构成 ρ-演算核心的底层形式语义。使用 ς-演算，我们可以将任何面向对象编程语言的语义1简化为四个元素：对象、方法、字段和类型。2由此产生的编程元素集足够小，我们可以提供它们可以组合和交互的方式的完整枚举。换句话说，我们可以描述面向对象编程中的任何关系，并且不能有其他关系。我们完全涵盖了可能的面向对象编程关系。
Although it may still seem self-referential, the formal reason for adding objects is that they form the heart and focal point of object-oriented programming. Without objects, there is no object-oriented programming. Not satisfying? Look to Abadi and Cardelli’s ς-calculus (sigma-calculus) instead. [1] ς-calculus is the underlying denotational semantics that forms the core of ρ-calculus. Using ς-calculus, we can reduce any object-oriented programming language’s semantics1 to the four elements: objects, methods, fields, and types.2 The resulting set of the elements of programming is sufficiently small that we can provide a complete enumeration of the ways in which they can combine and interact. In other words, we can describe any relationship in object-oriented programming, and there can be no others. We have a complete coverage of the possible object-oriented programming relationships.
从另一个角度来看，面向对象编程一直无法仅使用传统过程编程的方法（作为函数）、字段和类型来可靠且充分地建模。在尝试这样做时，会失去一些根本的东西。学者们已经就这一特定观点争论了三十年，但据我所知，每一次扩展过程编程底层 lambda 演算的语义以涵盖面向对象技术的努力都在某种程度上失败了。在每种情况下，无论过程扩展如何紧密地模拟面向对象系统的语义，接近所需的数学机制都非常繁琐、晦涩，并且无法与使用面向对象编程语言的概念优雅保持一致。另一方面，ς-calculus 通过将对象添加为第一类实体，成功地创建了一个数学模型，该模型反映了面向对象编程的心理模型，并提供了一种以正式方式处理对象的可靠而优雅的方法。
From another perspective, object-oriented programming has resisted being reliably and sufficiently modeled by using just the methods (as functions), fields, and types of traditional procedural programming. Something fundamental is lost in attempting to do so. Scholars have argued about this particular point for three decades now, but every attempt that I know of to extend the semantics of the lambda-calculus underlying procedural programming to encompass object-oriented techniques has fallen short in some way. In every case, regardless of how closely the procedural extensions model the semantics of object-oriented systems, the mathematical machinations required to come close are extremely cumbersome, obtuse, and fail to show any coherence with the conceptual elegance of using object-oriented programming languages. ς-calculus, on the other hand, by adding objects as first-class entities in their own right, successfully created a mathematical model that mirrors the mental model of object-oriented programming and gives a solid and elegant approach for working with objects in a formal manner.
A.1. 依赖运营商
A.1. Reliance Operators
ς-calculus 还扫除了除四个必需和必要的实体（对象、方法、字段和类型）之外的一切，为简单的分析开辟了道路真正全面的。例如，如果只给出四个项目，则只有 16 种组合方式。表 2.2说明了这一点，表 A.1中重现了这一点。
ς-calculus also swept aside everything but the four required and necessary entities—objects, method, fields, and types—opening the way for simple analyses that are truly comprehensive. For example, given only four items to work with, there are only sixteen ways they can be combined. This point was illustrated in Table 2.2, which is reproduced here in Table A.1.
Table A.1. All interactions between entities of object-oriented programming
该表很小，我们可以逐一分析。通过研究Defines ，我们可以一次性解决该表的大部分问题。对象在其内部定义方法、字段或类型。诸如命名空间或包对象之类的对象也可以在其中定义其他对象。在某些语言中，方法可以在其中定义内部方法和类型，但几乎总是可以定义对象和字段。字段本质上不容易包装或定义其他元素，但字段的类型（例如类）可以定义这四个元素中的任何一个。
This table is small enough that we can pick it apart piece by piece. We can address most of this table at one shot by investigating Defines. An object defines a method inside of it, or a field, or perhaps a type. Objects such as namespaces or package objects can also define other objects within them. Methods can, in some languages, define inner methods and types within them but can almost always define objects and fields. Fields, by their nature, do not readily wrap or define other elements, but the type of the field, such as a class, can define any of the four elements.
这种对内部实体的定义也称为作用域。定义方法的类就是在其自身内部限定该方法的作用域。必须通过类才能找到该方法，可以直接访问，也可以通过从该类实例化的对象或字段访问。对象限定其子级的作用域，方法也是如此。在 C++ 中，这些示例包括限定类中的静态方法或命名空间中的字段的作用域：Some-Class::aMethod()和ANamespace::aField。Java 避免使用双冒号作用域运算符，而是统一了范围符号，使用简单的点，如SomeClass.aMethod()或。这是 ς-calculus 使用的相同符号。在所有情况下，作用域都用点运算APackage.aField符 表示。对于大多数程序员来说，它应该看起来非常熟悉，因为它既用于限定绑定实例的作用域，也用于从定义中选择实体。在实践中真的没有什么区别。事实上，定义只是作用域的同义词。
This defining of inner entities is also known as scoping. A class that defines a method is scoping that method inside itself. You must go through the class to get to the method, either directly or through an object or field instantiated from that class. Objects scope their children, as do methods. Examples of these in C++ include scoping a static method in a class or a field in a namespace: Some-Class::aMethod() and ANamespace::aField respectively. Java eschews the double colon scoping operator and unifies the scoping notation across the board to use a simple dot, as in SomeClass.aMethod() or APackage.aField. This is the same notation that ς-calculus uses. In all cases, scoping is indicated by the dot operator. It should look very familiar to most programmers because it is used both to scope through bound instances and select entities out of a definition. There’s really no difference in practice. In fact, defines is just a synonym for scopes.
那么我为什么不直接使用Scopes呢？传统的过程式编程指称演算、λ 演算和 ς 演算之间的关键区别在于，从字段中选择元素会使用 σ 运算符。这就是将实例绑定到类型的原因，即使存在多态性，也可以在字段提供的范围之外正确查找所选项目。这种绑定是 ς-calculus 所独有的，再加上将对象视为一流实体，使得它极其有效和优雅。不过，出于本文的目的，我们可以忽略这种绑定。
So why didn’t I just use Scopes instead? The critical difference between the traditional denotational calculus of procedural programming, λ-calculus, and ς-calculus is that selecting an element from a field brings into play the σ operator. This is what binds an instance to a type, allowing for proper lookup of the selected item out of the scoping provided by the field, even when polymorphism is present. This binding is unique to ς-calculus and, combined with the treatment of objects as first-class entities, is what makes it extremely effective and elegant. We can, however, ignore this binding for the purposes of this text.
消除已知概念的定义后，我们得到表 2.3，再次在此处重现，即表 A.2。
Eliminating defines as a known concept leaves us with Table 2.3, again reproduced here, as Table A.2.
Table A.2. Nonscoping interactions between entities of object-oriented programming
对象和字段被声明为“属于某一类型”。ς-calculus 将其显示为anObject : itsType。此符号还用于指示方法的返回类型。类型互为子类型或继承。ς-calculus 将其显示为Subtype <: Supertype。3这样就消除了表 A.2中的全部内容，只留下四个依赖运算符：方法调用、字段使用、状态更改和内聚性。
Objects and fields are declared to be “of a type.” ς-calculus displays this as anObject: itsType. This notation is also used to indicate the return type of a method. Types subtype or inherit from one another. This is shown in ς-calculus as Subtype <: Supertype.3 That eliminates the entirety of Table A.2 except for the four reliance operators: method call, field use, state change, and cohesion.
基于上述四种关系，ρ 演算中有四种依赖运算符（简称 relops）。方法-方法 relop（即方法调用）是 mu 形式。这是本书中讨论的唯一依赖运算符。方法-字段（字段使用）是 phi 形式。字段-方法是一种状态变化（换句话说，字段的状态以某种方式依赖于方法调用），称为 sigma 形式。最后，字段-字段依赖是传统的内聚性或 kappa 形式。在 ρ 演算的数学中，它们分别表示为< μ、< φ、< σ和< κ。该符号来自子类型运算符< :，您可以将其视为“子类型的存在和意图依赖于超类型”。因为 : 是类型运算符，所以我们可以将<重新解释为“依赖”。然后，用适当的指示符为该基本符号添加下标，以形成四个依赖运算符。它们是二元运算符，从其使用的左侧到其右侧形成依赖。换句话说，someMethod < μ anotherMethod应该读作someMethod依赖于anotherMethod。
There are four reliance operators—relops for short—in ρ-calculus, based on the four relationships described above. The method-method relop (that is, a method call) is the mu-form. This is the single reliance operator discussed in this book. The method-field—a field use—is the phi-form. The field-method is a state change—in other words, the state of the field depends on a method call in some manner—known as the sigma-form. Finally, a field-field reliance is the traditional cohesion, or kappa-form. These are denoted in the mathematics of ρ-calculus as <μ, <φ, <σ, and <κ, respectively. The notation comes from the subtyping operator <:, which you can think of as stating, “The subtype relies on the supertype for its existence and intent.” Because : is a typing operator, we can reinterpret < as “relies on.” This basic notation is then subscripted with the appropriate indicator to form the four reliance operators. They are binary operators that form a reliance from the left-hand side of their use on their right-hand side. In other words, someMethod <μ anotherMethod should be read as someMethod relies on anotherMethod.
A.2. 及物性和同位素
A.2. Transitivity and Isotopes
在第 4.1.1 节中，您了解了同位素以及它们是如何通过间接依赖形成的。我们可以使用 ρ 演算通过传递性为这个过程提供形式基础。以下源示例重现了第 4.1.1 节中的代码示例。我们可以像以前一样声明f.foo依赖于b.bar，但现在我们可以将其写为f .foo < μ b.bar。到目前为止一切顺利。在右侧的示例中，我们可以说f .foo < μ g.goo并且g.goo < μ b.bar。
In Section 4.1.1, you learned about isotopes and how they are formed from indirect reliances. We can use ρ-calculus to give this process a formal basis through transitivity. The following source example reproduces our code example from Section 4.1.1. We can state, as before, that f.foo relies on b.bar, but now we can write this as f .foo <μ b.bar. So far, so good. In our example on the right side, we can say that f .foo <μ g.goo and that g.goo <μ b.bar.
我们可以在归约规则中使用这两条信息。这是 ς 演算和 ρ 演算中推理的主要形式。给定一组我们知道为真的子句，我们可以推断出一个新属性。我们说我们正在将这些子句归约到结果。例如，在这种情况下，对于< μ传递性有一个归约规则，如公式 A.1所示。
We can use these two bits of information in a reduction rule. This is the primary form of inference in ς-calculus and ρ-calculus. Given a set of clauses that we know to be true, we can infer a new property. We say that we are reducing the clauses to the result. For instance, in this case, there is a reduction rule for <μ transitivity, as in Equation A.1.
该归约规则首先指出我们有一个环境E，并且我们在此环境中定义了x、y和z。接下来的三个子句指出这些元素分别定义为具有子元素m、n和o，它们被指示为方法。接下来的两个子句指定了 relops，使得xm依赖于yn，而yn依赖于zo。给定这组子句，推断的结果是xm依赖于zo。这只是 ρ 演算中众多传递性之一，它们是是什么让 ρ 演算具有如此强的分析能力。极大的系统可以简化为简单的 relop 集，而且可以通过查看结果的简化视图来搜索高级抽象（如设计模式）的存在。如果没有这种技术，我们将不得不不断地处理或检查所有中间部分。例如，如果我们要寻找xm调用zo的证据，那么对这个系统的简单检查将无法找到它。通过使用 ρ 演算的传递性并声明我们正在寻找xm依赖于zo 的情况，我们可以快速而干净地证明这种关系的存在。
This reduction rule begins by stating that we have an environment E such that we have defined in this environment x, y, and z. The next three clauses state that these elements are defined as having subelements m, n, and o, respectively, which are indicated to be methods. The next two clauses specify relops such that x.m relies on y.n, and y.n relies on z.o. Given this set of clauses, the inferred result is that x.m relies on z.o. This is just one of the many transitivities in ρ-calculus, and they are what give ρ-calculus much of its analytic strength. Extremely large systems can be reduced to simple sets of relops, and the existence of high-level abstractions such as design patterns can be searched for by looking in the simplified views of the results. Without this technique, we would have to continually work with or examine all of the intervening pieces. If, for example, we were looking for proof that x.m called z.o, a simple examination of this system would fail to find it. By using the transitivities of ρ-calculus and stating that we are looking for a situation where x.m relies on z.o instead, we can prove the existence of the relationship quickly and cleanly.
更确切地说，这种传递性使我们能够为设计模式创建特定的基于依赖的定义，这些定义可以匹配大量结构实现。这些是规范、最基本和最直接的实现的同位素。这就是这种方法如此强大和灵活的原因。
More precisely, this transitivity allows us to create specific reliance-based definitions for design patterns that can match a tremendous number of structural implementations. These are the isotopes of the canonical, most basic and direct implementation. This is what makes this approach so powerful and flexible.
A.3. 相似性
A.3. Similarity
同样，第 2.2.2 节中的三个相似性轴也可以在 ρ 演算中得到形式化基础。简单提醒一下，单个 relop 的两侧之间有三种可能相似性。例如，给定依赖f.foo < μ b.bar ，我们可以讨论f和b之间的对象相似性、 f和b的类型之间的相似性以及方法foo和bar之间的相似性。在 ρ 演算中，我们使用 ~ 表示相似性，使用≁表示不相似性。例如，A ~ B表示“ A与B相似”，B ≁ C表示“ B与C不相似”。
Our three axes of similarity from Section 2.2.2 can likewise be given a formal basis in ρ-calculus. As a quick reminder, there are three possible ways in which a single relop can have similarities between its two sides. For instance, given the reliance f.foo <μ b.bar, we can talk about the object similarity between f and b, the similarity between the types of f and b, and the similarity between the methods foo and bar. We use ~ to indicate similarity in ρ-calculus and ≁ for dissimilarity. For instance, A ~ B for “A is similar to B” or B ≁ C for “B is dissimilar to C.”
类型之间的相似性可以在 ρ 演算的f和b定义中找到。为了在 ρ 演算定义中使用f和b，必须定义它们，并且必须为它们指定类型。因此，f和b的类型信息必须已经定义。但是，我们可以将这种类型关系视为以下之一：不相关，我们称之为不相似，并用≁表示；完全相同的类型，我们称之为相似并标记为 ~；子类型，用之前看到的< : 表示；以及兄弟类型，其中两种类型共享一个公共超类型，用< : >表示。所有这些信息都已存在于 ρ 演算中。
The similarity between the types is found in the ρ-calculus definition of f and b. For us to use f and b in a ρ-calculus definition, they must be defined, and they must be given types. Therefore, the typing information of f and b must already be defined. We can talk about this typing relationship, however, as one of the following: unrelated, which we will just call dissimilar, and denote by ≁; the same exact type, which we call similar and mark as ~; subtyping, denoted by the previously seen <:; and sibling typing, in which two types share a common supertype, indicated by <:>. All of this information is already present in ρ-calculus.
我们可以使用相同的 ~ / ≁符号来讨论对象和方法之间的相似性，即相似或不相似。但为了简洁起见，并且由于此信息对于特定的依赖运算符实例是唯一的，我们可以将此信息直接作为上标添加到依赖运算符符号中，尽管形式略有修改。
We can talk about the similarity between objects and methods as either similar or dissimilar as well, using the same ~/≁ notation. For brevity, however, and because this information is unique to specific reliance operator instances, we can add this information to the reliance operator notation directly as a superscript, albeit in a slightly modified form.
虽然 ~/ ≁符号在正常文本大小下很容易区分，但在小字体下会出现可读性问题，因此为了清晰起见，我们改用 +/– 符号。4 点运算符的选择符号在相似性符号中得到模仿：对象相似性放在点的左侧，方法相似性放在右侧。例如，具有对象差异性和方法相似性的方法调用 relop 将被记为
。具有对象相似性和方法差异性的 relop 将显示为
，等等。这使我们能够将大量信息压缩成简洁的形式。结合传递性，它使我们能够快速、准确、清晰地描述系统中任何两个实体之间的关系。
Although the ~/≁ notation is easily distinguishable at normal text sizes, at small font sizes it presents readability problems, so +/– is used instead for clarity.4 The selection notation of the dot operator is mimicked in the similarity notation: the object similarity is placed on the left of a dot and the method similarity is placed on the right. For instance, a method call relop with object dissimilarity and method similarity would be noted as . A relop with object similarity and method dissimilarity would be shown as , and so on. This lets us collapse a tremendous amount of information into a concise form. Combined with the transitivities, it lets us describe relationships between any two entities in a system quickly, precisely, and cleanly.
你可能会想到，我们可能并不总是知道系统中两个元素之间的相似性或不相似性。你是对的。在这种情况下，圆圈 ο 用于表达真正未知的关系。
It might occur to you that we may not always know the similarity or dissimilarity between two elements of a system. You are correct. In such cases, a circle, ο, is used to convey a truly unknown relationship.
A.4. EDP 形式化
A.4. EDP Formalisms
现在，我们有了正式定义和使用 ρ 演算来为 EDP 奠定基础的基础。我们可以重新查看第 4.4 节中的 EDP 设计空间图，并使用我们的新符号重新注释它们，如图A.1和A.2所示。
We now have the basis for formally defining and working with ρ-calculus to create the foundation for the EDPs. We can revisit the EDP design space diagrams from Section 4.4 and reannotate them with our new notation, as in Figures A.1 and A.2.
图 A.1. 完整的方法调用 EDP 设计空间：类似方法。
Figure A.1. The full method call EDP design space: similar method.
图 A.2. 完整方法调用 EDP 设计空间：不同方法。
Figure A.2. The full method call EDP design space: dissimilar method.
让我们为这个讨论建立一个简单的设置。假设我们有一个类型A的对象o和一个类型B的对象p。我们可以分别将它们写为o : A和p : B。类型A定义一个方法f，类型B定义一个方法g。此设置可以在代码中显示，如清单 A.1所示。
Let’s establish a simple setup for this discussion. Assume we have an object o of type A and an object p of type B. We can write these as o : A and p : B, respectively. Type A defines a method f and type B defines a method g. This setup could be shown in code something like in Listing A.1.
我们现在可以使用相似性符号和依赖运算符，根据 EDP 在设计空间中的位置来定义 EDP。
We can now define the EDPs on the basis of their placement within the design space, using our similarity notation and reliance operators.
从图 A.1可以看出，递归位于对象、类型和方法相似性轴的交汇处。这意味着递归可以正式表示为pg 
, A ~ B。回想一下，上标表示匹配范围点运算符两侧的两个元素之间的关系。在本例中，o与p相似，f与g相似。
Looking at Figure A.1, we can see that Recursion is at the intersection of the axes of object, type, and method similarity. This means that Recursion can be expressed formally as o.f p.g, A ~ B. Recall that the superscript indicates what the relationship is between the two elements on either side of the matching scoping dot operator. In this case, that o is similar to p and f is similar to g.
你可以看到，委托可以写为pg 
， A≁B ， 并且，正如你所料，重定向可以写为pg 。第 2.2.4 节中的声明 因此，从第 17 页开始，重定向和委托共同构成了耦合的概念。我们的“未知”符号终于开始发挥作用，尽管作用很小。 
You can see that Delegation can be written as o.f p.g, A ≁ B, and, as you might expect, Redirection is written o.f p.g. The claim from Section 2.2.4 that Redirection and Delegation together form the concept of coupling can be stated, therefore, as o.f p.g. Our “unknown” notation finally comes into play, albeit in a minor way.
Listing A.1. Simple code example
类A {
2 void f();
};
4
类B {
6 void g();
};
8
A o;
10 B p;
class A {
2 void f();
};
4
class B {
6 void g();
};
8
A o;
10 B p;
我很确定您现在可以猜出Conglomeration在这一点上是什么样子的：pg的，现在 第 2.2.4 节中讨论的凝聚力减少到pg的，因为来自递归的相似性和来自Conglomeration 的不同性相互抵消。 
I’m pretty sure you can guess what Conglomeration looks like at this point: o.f p.g, and now the cohesion discussed in Section 2.2.4 reduces to o.f p.g, as the similarity from Recursion and the dissimilarity from Conglomeration cancel each other out.
对于Redirected Recursion，方法依赖就是你对pg的Redirection :所期望的。顺便说一句，这是我们采用Redirected Recursion作为名称的正式理由 - 这是Redirection的改进，而不是Recursion。因此，只需要将类型关系更改为A = B。
For Redirected Recursion, the method reliance is what you’d expect for a Redirection: o.f p.g. By the way, this is our formal reasoning for going with Redirected Recursion as a name—this is a refinement of Redirection, not Recursion. As a result, only the type relationship needs to be changed, to A = B.
对于图 A.1底部的最后两个 EDP ，我们添加了子类型和兄弟关系。同样，方法依赖关系正是pg的重定向所期望的，我们再次只更改类型关系。这次它设置为A < : B，这表示A是B的子类型，从而产生了受信任的重定向EDP。
For the final two EDPs along the bottom row of Figure A.1, we add our subtyping and sibling relationships. Again, the method reliance is what you’d expect for a Redirection, o.f p.g, and once more we only change the type relation. This time it is set to A <: B, which indicates that A is a subtype of B, giving rise to the Trusted Redirection EDP.
我们可以再次从 重定向 开始，从而实现代理重定向，这次对类型依赖关系进行简单的更改，例如A < : > B。
We can arrive at Deputized Redirection by again starting with Redirection, this time making a simple change to the type reliance relationship, such that A <:> B.
最后，我们重新回顾一下Recursion。从它的调用依赖运算符和类型语句开始：of pg，o : A，p : B。将类型关系更改为A < : B。这就是修改Recursion以将其演变为Extend Method 的全部内容。
Finally, let’s revisit Recursion. Start with its call reliance operator and type statements: o.f p.g, o : A, p : B. Change the type relationship to A <: B. That’s all there is to modifying Recursion to evolve it into Extend Method.
我们可以使用这种方法来定义此设计空间中的任何 EDP，并使用 ρ 演算优雅而简洁地表达它们的核心关系。第 A.7 节使用来自第 A.2 节的归约规则符号为每个 EDP 提供了完整定义。给定适当的子句集，我们可以推断归约规则底部的事实是显而易见的，进而将其用作其他归约规则推理中的新子句。例如，以递归的定义为例：
We can use this approach to define any of the EDPs within this design space, elegantly and concisely expressing their core relationships using the ρ-calculus. Section A.7 provides full definitions for each EDP using our reduction rule notation from Section A.2. Given the proper set of clauses, we can infer that the fact at the bottom of a reduction rule is evident and in turn use it as a new clause in other reduction rule inferences. For instance, take the definition for Recursion:
上面的子句与我们定义Recursion时看到的类似。第一个子句指出Recursor是一种类型，它至少有一个名为operation 的方法，还有其他可选的方法和/或字段。第二个子句确定对象r属于Recursor类型。第三个子句指出r.operation具有对自身的方法调用依赖，具有对象和方法的相似性。最后一行指出，前面子句中的元素构成递归。我们使用正式定义中的名称作为模式角色的名称。这些名称既用于讨论模式，也用于定义其等效 PIN 表示的结构，如您将在A.6 节中看到的那样。与这些角色绑定的槽是数学意义上的变量。一旦实体与角色绑定，该角色就被认为由该实体履行。在前面的例子中，任何满足定义中所述的递归需求的类都被称为与递归的 Recursionor 角色绑定。
The clauses above the line are analogous to what we saw when we defined Recursion. The first clause states that Recursor is a type and that it has at least one method, which is named operation, with other optional methods and/or fields. The second clause establishes that the object r is of type Recursor. The third clause states that r.operation has a method call reliance on itself, with both object and method similarity. The final line states that the elements in the prior clauses form an instance of Recursion. We use the names in the formal definition as the names of the roles for the pattern. These are used both when discussing the pattern and for defining the structure of its equivalent PIN representation, as you’ll see in Section A.6. The slots that are bound to these roles are variables in the mathematical sense. Once an entity is bound to a role, that role is said to be fulfilled by that entity. In the preceding case, any class that satisfies the needs of Recursion as stated in the definition is said to be bound to the Recursor role of Recursion.
A.5. 组合和缩减规则
A.5. Composition and Reduction Rules
从这些定义开始，我们可以将它们组合成更复杂的模式，就像我们在前面的章节中所做的那样。让我们重新审视一下第 4.2 节中构建的Decorator。回顾一下，我们引入了Fulfill Method作为Inheritance和Abstract Method的组合，然后使用Fulfill Method定义Objectifier。我们结合Objectifier和Trusted Redirection来定义Object Recursion，然后将其与Extend Method结合起来创建了Decorator。
Starting with these definitions, we can compose them into more complex patterns, just like we did in the earlier chapters. Let’s reexamine the building of Decorator that we undertook in Section 4.2. To recap, we introduced Fulfill Method as a combination of Inheritance and Abstract Method, and then used Fulfill Method to define Objectifier. We combined Objectifier and Trusted Redirection to define Object Recursion, which was then brought together with Extend Method to create Decorator.
让我们首先将Fulfill 方法定义为：
Let’s start by defining Fulfill Method as:
第一个子句声明了一个Abstract Interface实例，该实例接受两个参数，第一个是类型实体，第二个是方法实体。它按照Abstract Interface模式中的定义建立了它们之间的关系，但是允许我们用这个较短的版本替换完整的形式定义。第二个子句通过Inheritance EDP的实例建立了ConcreteClass是Abstractor的子类。第三行是谜题的新部分。它指出Abstract Interface中声明的方法由ConcreteClass定义，履行了Abstract Interface做出的承诺。请注意，这一部分至关重要，因为不需要子类来实际定义抽象方法——它可能只是类层次结构中的另一个抽象类。子句中提供的名称并非偶然。如果一个名称在缩减规则中出现在多个位置，则系统中的每个角色都使用相同的实体。这就是将较小的模式联系成一个更大的组合的原因。
The first clause declares an instance of Abstract Interface, which takes two arguments, the first a type entity, and the second a method entity. It establishes the relationship between them as defined in the Abstract Interface pattern but allows us to replace the full formal definition with this shorter version. The second clause establishes that ConcreteClass is a subclass of Abstractor through an instance of the Inheritance EDP. The third line is the new piece of the puzzle. It states that the method stated in Abstract Interface is defined by ConcreteClass, fulfilling the promise made by Abstract Interface. Note that this piece is crucial, as there is no need for a subclass to actually define the abstract method—it may just be another abstract class in the class hierarchy. The names provided in the clauses are not accidental. If a name appears in more than one position in a reduction rule, then the same entity in the system is being used in each role. This is what ties together the smaller patterns into a larger composition.
我们可以继续简化规则的过程，并通过向多个操作添加一个 Client 类来定义Objectifier来抽象：
We can continue the reduction rules process and define Objectifier by adding a Client class to a plurality of operations to abstract:
受信任的重定向看起来很像我们之前对递归的定义：
Trusted Redirection looks rather like our earlier definition for Recursion:
对象递归的正式定义是：
The formal definition of Object Recursion is then:
我们正在确定一些Objectifier ConcreteClasses 参与了受信任的重定向，而另一些则不参与。如果所有这些类都参与了重定向，则对该架构的调用将永远不会终止。
We’re establishing that some Objectifier ConcreteClasses are involved in Trusted Redirection and some are not. If all such classes were involved in redirection, a call into this architecture would never terminate.
接下来，我们可以使用简写符号来建立元素是类的方法，如aMethod ∊ meth ( aClass )，并将Extend Method定义为：
Next, we can use a shorthand notation for establishing that an element is a method of a class as in aMethod ∊ meth(aClass) and define Extend Method as:
最后，我们可以将Decorator定义为：
Finally, we can define Decorator as:
请注意，这些方程式可以与第 4.2 节中用于说明组合的 PIN 图密切相关。事实上，正如我们接下来将看到的，PIN 和简化规则中模式实例的使用是相辅相成的。
Note that these equations can be correlated closely with the PIN diagrams used in Section 4.2 to illustrate the compositions. In fact, as we’ll see next, PIN and the uses of pattern instances in reduction rules go hand in hand.
A.6. 模式实例符号和角色
A.6. Pattern Instance Notation and Roles
PIN 的一个设计目标是允许通过扩展的 PIN 框简单地显示组合模式。PIN 的规则与本附录中缩减规则组合的规则非常相似。例如，图 A.3重复了图 3.9，在 ρ 演算的形式化中，这些模式的实例将表现为PatternA（Role1 : a, Role2 : b）和PatternB（Role1 : a, Role2 : b, Role3 : c, Role4 : d, Role5 : e, Role6 : f），其中a到f代表未绑定的角色。
One design goal of PIN is to allow the simple display of composed patterns through expanded PINboxes. The rules for PIN closely mirror those for the reduction rule compositions in this appendix. For instance, Figure A.3 is a repeat of Figure 3.9, and instances of those patterns in the formalism of ρ-calculus would appear as PatternA(Role1 : a, Role2 : b) and PatternB(Role1 : a, Role2 : b, Role3 : c, Role4 : d, Role5 : e, Role6 : f), where a through f represent unbound roles.
图 A.3. 标准 PINbox。
Figure A.3. Standard PINbox.
同样，图 A.4展示了图 3.14中的扩展 PINbox 示例，我们现在可以将其正式定义为：
Likewise, Figure A.4 shows our expanded PINbox example from Figure 3.14, which we can now define formally as:
图 A.4. 扩展的 PIN 实例。
Figure A.4. Expanded PIN instance.
PINbox 中的角色与正式定义中的角色之间存在一一对应关系。此外，PINbox 图中的每个连接都说明了在定义中的多个角色履行中使用同一绑定变量的点。
There is a one-to-one correlation between the roles ringing a PINbox and the roles in the formal definition. Also, each connection in the PINbox diagram illustrates a point at which the same bound variable is used in multiple role fulfillments within the definition.
A.7. EDP 定义
A.7. EDP Definitions
本节使用 ρ 演算对本文中的每个 EDP 提供了完全形式化的定义。虽然您已经看到了所使用的大多数符号，但并未完整解释这些定义的每个方面。如果您希望进行更深入的了解，请参阅基础出版物以获得完整的参考资料，特别是《对象理论》 [ 1 ]，它为您提供了所需的全面基础，以及《SPQR：从源代码自动检测设计模式的形式基础和实践支持》 [ 35 ]，它提供了原始定义以及对每个定义的更深入的讨论。自最初出版以来，一些小细节发生了变化，例如在模式实例中增加了对角色的明确命名。在原始参考资料发表后，发现这非常有用。以下每个定义都附有其等效的 PIN 图，因此您可以立即看到与形式主义的相关性。
This section provides fully formal definitions of each of the EDPs in this text, using ρ-calculus. Not every aspect of these definitions is explained in full, although you have seen most of the notation used. If you would like a deeper treatment, refer to the base publications for the complete reference, in particular A Theory of Objects [1], which provides the comprehensive foundation you need, as well as SPQR: Formal Foundations and Practical Support for the Automated Detection of Design Patterns from Source Code [35], which provides the original definitions and a much deeper discussion of each. Some minor details have changed since original publication, such as the addition of the explicit naming of the roles in pattern instances. This was found to be of great utility after the original references were published. Each of the following definitions is accompanied by its equivalent PIN diagram so you can immediately see the correlation with the formalism.
A.7.1. 创建对象
A.7.1. Create Object
此定义最接近 ς-calculus 中的基础，如果您希望真正详细了解此定义，建议您研究该基础。或者，您可以接受这是针对一个非常简单的概念的非常具体的定义。
This definition is the closest to its foundations in ς-calculus, and studying that foundation is recommended if you wish to truly understand this definition in detail. Or, you can accept that this is a very specific definition for a very simple concept.
其中
（并且 可能是Root），根据 [ 1 ]中建立的分类，分别对于 O-1 有⌊ A ⌋ = A ，对于 O-2 有X < : A ，对于符合 O-3 的语言有X < # A。
where (and may be Root), ⌊A⌋ = A for O-1, X <: A for O-2, and X < #A for O-3-compliant languages, respectively, according to the classifications established in [1].
请注意，将其视为语言模型直接输出的公理可能比尝试从 ς-calculus 构造中推导出来更容易。特别是，如果尝试执行源代码分析，请使用语言语义，而不是尝试从 ς-calculus 原语中推导出来。
Note that it may very well be easier to consider this as an axiomatic output directly from a language model than to try to derive it from ς-calculus constructs. In particular, if attempting to perform analysis of source code, use the language semantics instead of trying to derive this from ς-calculus primitives.
A.7.2. 检索
A.7.2. Retrieve
大多数面向对象语言不允许用新值（方法体）更新方法。相反，只能以这种方式更新数据字段。从理论角度来看，这两种情况没有明显的区别。我们允许这两种情况，依靠语言语义来控制哪些情况有效，哪些情况无效。通过推迟我们的定义，直到我们有一组格式良好的 ρ 演算事实，我们避免了处理大量语言怪癖的大部分复杂性。检索模式有两种基本形式：
Most object-oriented languages do not allow the updating of methods with new values (method bodies). Instead, only data fields can be updated in this manner. From a theoretical point of view, there is no appreciable difference between these two scenarios. We allow for both, relying on language semantics to govern which cases are valid and which are not. By deferring our definition until we have a set of well-formed ρ-calculus facts, we avoid much of the complexity of handling the myriad of language quirks. There are two basic forms for the Retrieve pattern:
两种形式均由第一个子句中的更新和第二个子句中的返回值组成，返回值可以按名称（表示共享引用）或按值（表示新副本）。
Both forms are composed of an update in the first clause and a return of a value in the second clause, either by name, indicating a shared reference, or by value, indicating a fresh copy.
A.7.3. 继承
A.7.3. Inheritance
A.7.4. 抽象接口
A.7.4. Abstract Interface
A.7.5. 委托
A.7.5. Delegation
A.7.6. 重定向
A.7.6. Redirection
A.7.7. 聚集
A.7.7. Conglomeration
A.7.8. 递归
A.7.8. Recursion
A.7.9. 恢复方法
A.7.9. Revert Method
A.7.10. 扩展方法
A.7.10. Extend Method
A.7.11. 委托聚合
A.7.11. Delegated Conglomeration
A.7.12. 重定向递归
A.7.12. Redirected Recursion
A.7.13. 受信任的委托
A.7.13. Trusted Delegation
A.7.14. 受信任的重定向
A.7.14. Trusted Redirection
A.7.15. 代理代表团
A.7.15. Deputized Delegation
A.7.16. 代理重定向
A.7.16. Deputized Redirection
A.8. 中间模式定义
A.8. Intermediate Pattern Definitions
您将在这里找到第 6 章中介绍的模式的定义。这些是初始模式，展示了先前模式实例的组合缩减规则。同样，每个定义都与其等效的 PIN 图一起呈现。
Here you will find the definitions for the patterns presented in Chapter 6. These are the initial patterns demonstrating composition reduction rules from prior pattern instances. Again, each definition is presented along side its equivalent PIN diagram.
A.8.1. Fulfill 方法
A.8.1. Fulfill Method
A.8.2. 检索新
A.8.2. Retrieve New
A.8.3. 检索共享
A.8.3. Retrieve Shared
A.8.4. 对象化器
A.8.4. Objectifier
A.8.5. 对象递归
A.8.5. Object Recursion
A.9. 四人帮模式定义
A.9. Gang of Four Pattern Definitions
最后，本节介绍第 7 章中讨论的四人帮模式的定义和 PIN 图。
Finally, this section presents definitions and PIN diagrams for each of the Gang of Four patterns discussed in Chapter 7.
A.9.1. 抽象工厂
A.9.1. Abstract Factory
A.9.2. 工厂方法
A.9.2. Factory Method
A.9.3. 装饰器
A.9.3. Decorator
A.9.4. 代理
A.9.4. Proxy
A.9.5. 责任链
A.9.5. Chain of Responsibility
A.9.6. 模板方法
A.9.6. Template Method
参考书目
Bibliography
[1] 马丁·阿巴迪和卢卡·卡德利。对象理论。施普林格出版社，纽约，1996 年。
[1] Martín Abadi and Luca Cardelli. A Theory of Objects. Springer-Verlag, New York, 1996.
[2] Lee Ackerman 和 Celso Gonzalez。基于模式的工程：通过模式成功交付解决方案。Addison-Wesley，波士顿，2010 年。
[2] Lee Ackerman and Celso Gonzalez. Patterns-Based Engineering: Successfully Delivering Solutions via Patterns. Addison-Wesley, Boston, 2010.
[3] Christopher Alexander、Sara Ishikawa 和 Murray Silverstein。《模式语言：城镇、建筑和施工》。牛津大学出版社，纽约，1977 年。
[3] Christopher Alexander, Sara Ishikawa, and Murray Silverstein. A Pattern Language: Towns, Building, Construction. Oxford University Press, New York, 1977.
[4] Christopher W. Alexander. 《形式综合笔记》（第 15 版）。牛津大学出版社，纽约，1964 年，1999 年。
[4] Christopher W. Alexander. Notes on the Synthesis of Form (15th printing). Oxford University Press, New York, 1964, 1999.
[5] Kent Beck。Smalltalk 最佳实践模式。 Prentice Hall，Upper Saddle River，新泽西州，1997 年。
[5] Kent Beck. Smalltalk Best Practice Patterns. Prentice Hall, Upper Saddle River, NJ, 1997.
[6] Pete Becker。编程语言 c++ 标准工作草案。技术报告 N3242，ISO/IEC JTC/SC22/工作组 21，2011 年。
[6] Pete Becker. Working draft, standard for programming language c++. Technical Report N3242, ISO/IEC JTC/SC22/Working Group 21, 2011.
[7] James M. Bieman 和 Byung-Kyoo Kang。面向对象系统中的内聚性和重用。载于ACM 软件可重用性研讨会论文集 SSR'95，第 259-262 页，1995 年 4 月。重印于 ACM 软件工程笔记，1995 年 8 月。
[7] James M. Bieman and Byung-Kyoo Kang. Cohesion and reuse in an object-oriented system. In Proceedings of the ACM Symposium on Software Reusability, SSR’95, pp. 259–262, Apr 1995. Reprinted in ACM Software Engineering Notes, Aug 1995.
[8] James M. Bieman 和 Byung-Kyoo Kang。测量设计层面的凝聚力。IEEE软件工程学报，24(2):111–124，1998。
[8] James M. Bieman and Byung-Kyoo Kang. Measuring design-level cohesion. IEEE Transactions on Software Engineering, 24(2):111–124, 1998.
[9] James M. Bieman 和 Linda Ott。《功能内聚性的测量》。《IEEE 软件工程学报》，20(8):644–657，1994 年。
[9] James M. Bieman and Linda Ott. Measuring functional cohesion. IEEE Transactions on Software Engineering, 20(8):644–657, 1994.
[10] Grady Booch。部落记忆。IEEE软件，25(2):16-17，2008。
[10] Grady Booch. Tribal memory. IEEE Software, 25(2):16–17, 2008.
[11] Grady Booch 和 Celso Gonzalez。软件架构手册。www.handbookofsoftwarearchitecture.com/ ， 2010 年 10 月。
[11] Grady Booch and Celso Gonzalez. Handbook of software architecture. www.handbookofsoftwarearchitecture.com/, Oct 2010.
[12] Jan Bosch。设计模式作为语言构造。面向对象编程杂志，1(2):18-52，1998 年 5 月。
[12] Jan Bosch. Design patterns as language constructs. Journal of Object Oriented Programming, 1(2):18–52, May 1998.
[13] LC Briand 和 JW Daly。面向对象系统中内聚性度量的统一框架。《METRICS'97 第四届大会论文集》，第 43-53 页，1997 年 11 月。
[13] L. C. Briand and J. W. Daly. A unified framework for cohesion measurement in object-oriented systems. In Proceedings of the Fourth Conference on METRICS’97, pp. 43–53, Nov 1997.
[14] William J. Brown、Raphael C. Malveau、William H. Brown、W. McCormick Hays III 和 Thomas J. Mowbray。《反模式：危机中的重构软件、架构和项目》。Wiley，纽约，1998 年。
[14] William J. Brown, Raphael C. Malveau, William H. Brown, W. McCormick Hays III, and Thomas J. Mowbray. Anti-Patterns: Refactoring Software, Architectures, and Projects in Crisis. Wiley, New York, 1998.
[15] Shyam R. Chidamber 和 Chris F. Kemerer。面向对象设计的度量套件。《OOPSLA '91 会议纪要》，第 197-211 页。ACM，1991 年。
[15] Shyam R. Chidamber and Chris F. Kemerer. Towards a metrics suite for object-oriented design. In Proceedings of OOPSLA ’91, pp. 197–211. ACM, 1991.
[16] Shyam R. Chidamber 和 Chris F. Kemerer。面向对象设计的度量套件。IEEE软件工程学报，20(6):476–493，1994 年。
[16] Shyam R. Chidamber and Chris F. Kemerer. A metrics suite for object-oriented design. IEEE Transactions on Software Engineering, 20(6):476–493, 1994.
[17] James Coplien。C++ 习语。《第三届欧洲编程和计算模式语言会议论文集》，1998 年。
[17] James Coplien. C++ idioms. In Proceedings of the Third European Conference on Pattern Languages of Programming and Computing, 1998.
[18] C++ 标准委员会。工作草案，编程语言 C++ 标准，第 20.7.2.2 节 [util.smartptr.shared]。ISO/IEC 文档 N3242=11-0012，2011 年 2 月。
[18] C++ Standards Committee. Working Draft, Standard for Programming Language C++, Section 20.7.2.2 [util.smartptr.shared]. ISO/IEC Document N3242=11-0012, Feb 2011.
[19] Martin Fowler。重构：改进现有代码的设计。Addison-Wesley，波士顿，1999 年。
[19] Martin Fowler. Refactoring: Improving the Design of Existing Code. Addison-Wesley, Boston, 1999.
[20] Martin Fowler 和 Kendall Scott。UML Distilled：标准对象建模语言简介，第 3 版。Addison-Wesley，波士顿，2003 年。
[20] Martin Fowler and Kendall Scott. UML Distilled: A Brief Guide to the Standard Object Modeling Language, 3rd ed. Addison-Wesley, Boston, 2003.
[21] Erich Gamma、Richard Helm、Ralph Johnson 和 John Vlissides。《设计模式》。Addison-Wesley，波士顿，1995 年。
[21] Erich Gamma, Richard Helm, Ralph Johnson, and John Vlissides. Design Patterns. Addison-Wesley, Boston, 1995.
[22] Byung-Kyoo Kang 和 James M. Bieman。设计层面的凝聚力测量：推导、比较和应用。第 20 届国际计算机软件和应用程序会议 (COMPSAC'96) 论文集，第 92-97 页，1996 年 8 月。
[22] Byung-Kyoo Kang and James M. Bieman. Design-level cohesion measures: Derivation, comparison, and applications. In Proceedings of the 20th International Computer Software and Applications Conference (COMPSAC’96), pp. 92–97, Aug 1996.
[23] Byung-Kyoo Kang 和 James M. Bieman。利用设计凝聚力来可视化、量化和重构软件。第八届软件工程和知识工程国际会议，SEKE '96，1996 年 6 月。
[23] Byung-Kyoo Kang and James M. Bieman. Using design cohesion to visualize, quantify and restructure software. In Eighth International Conference of Software Engineering and Knowledge Engineering, SEKE ’96, Jun 1996.
[24] Joshua Kerievsky。《重构模式》。Addison-Wesley，波士顿，2005 年。
[24] Joshua Kerievsky. Refactoring to Patterns. Addison-Wesley, Boston, 2005.
[25] Howard C. Lovatt、Anthony M. Sloane 和 Dominic R. Verity。Java 的模式强制编译器 (PEC)：使用编译器。收录于 Sven Hartmann 和 Markus Stumptner 编的《信息技术研究与实践会议》第 43 卷。发表于 2005 年第二届亚太概念建模会议 (APCCM2005)。
[25] Howard C. Lovatt, Anthony M. Sloane, and Dominic R. Verity. A pattern enforcing compiler (PEC) for Java: Using the compiler. In Sven Hartmann and Markus Stumptner, eds., Conferences in Research and Practice in Information Technology, vol. 43. Appeared at the Second Asia-Pacific Conference on Conceptual Modeling (APCCM2005), 2005.
[26] OL Madsen、B. Møller-Pederson 和 K. Nygaard。BETA语言中的面向对象编程。Addison-Wesley，波士顿，1993 年。
[26] O. L. Madsen, B. Møller-Pederson, and K. Nygaard. Object-Oriented Programming in the BETA Language. Addison-Wesley, Boston, 1993.
[27] Robert C. Martin。《代码整洁之道：敏捷软件工艺手册》。Prentice Hall PTR，Upper Saddle River，新泽西州，2008 年。
[27] Robert C. Martin. Clean Code: A Handbook of Agile Software Craftsmanship. Prentice Hall PTR, Upper Saddle River, NJ, 2008.
[28] Scott Meyers。Effective C++。Addison-Wesley，波士顿，1992 年。
[28] Scott Meyers. Effective C++. Addison-Wesley, Boston, 1992.
[29] Scott Meyers。更有效的 C++。Addison-Wesley，波士顿，1996 年。
[29] Scott Meyers. More Effective C++. Addison-Wesley, Boston, 1996.
[30] Jörg Niere、Lothar Wendehals 和 Albert Zündorf。一种交互式可扩展的设计模式恢复方法。技术报告 tr-ri-03-236，帕德博恩大学，德国帕德博恩，2003 年 1 月。
[30] Jörg Niere, Lothar Wendehals, and Albert Zündorf. An interactive and scalable approach to design pattern recovery. Technical Report tr-ri-03-236, University of Paderborn, Paderborn, Germany, Jan 2003.
[31] S. Patel、W. Chu 和 R. Baxter。复合模块内聚性的度量。国际软件工程大会，第 38-48 页，1992 年 5 月。
[31] S. Patel, W. Chu, and R. Baxter. A measure for composite module cohesion. In International Conference on Software Engineering, pp. 38–48, May 1992.
[32] Dirk Riehle。复合设计模式。1997年 ACM SIGPLAN 面向对象编程系统、语言和应用程序会议论文集，第 218-228 页。ACM 出版社，1997 年。
[32] Dirk Riehle. Composite design patterns. In Proceedings of the 1997 ACM SIGPLAN Conference on Object-Oriented Programming Systems, Languages and Applications, pp. 218–228. ACM Press, 1997.
[33] James Rumbaugh、Ivar Jacobson 和 Grady Booch。《统一建模语言参考手册》，第 2 版，Addison-Wesley，波士顿，2004 年。
[33] James Rumbaugh, Ivar Jacobson, and Grady Booch. The Unified Modeling Language Reference Manual, 2nd ed., Addison-Wesley, Boston, 2004.
[34] MH Samadzadeh 和 SJ Khan。面向对象软件系统的稳定性、耦合和内聚性。第 22 届 ACM 计算机科学年度会议论文集，第 312-319 页，1994 年 3 月 8-10 日。
[34] M. H. Samadzadeh and S. J. Khan. Stability, coupling and cohesion of object-oriented software systems. In Proceedings of the 22nd Annual ACM Computer Science Conference on Scaling Up, pp. 312–319, Mar 8–10, 1994.
[35] Jason McC. Smith。SPQR ：从源代码自动检测设计模式的形式基础和实际支持。博士论文，北卡罗来纳大学教堂山分校，2005 年 12 月。
[35] Jason McC. Smith. SPQR: Formal Foundations and Practical Support for the Automated Detection of Design Patterns from Source Code. Ph.D. thesis, University of North Carolina at Chapel Hill, Dec 2005.
[36] Jason McC. Smith。模式实例符号：一种用于动态可视化和理解软件模式的简单分层可视化符号。《视觉语言与计算杂志》，22(5):355–374，2011 年。
[36] Jason McC. Smith. The Pattern Instance Notation: A simple hierarchical visual notation for the dynamic visualization and comprehension of software patterns. Journal of Visual Languages and Computing, 22(5):355–374, 2011.
[37] Jason McC。Smith 和 David Stotts。SPQR：从源代码灵活自动提取设计模式。第 18 届 IEEE 国际自动软件工程大会，第 215-224 页，2003 年 10 月。
[37] Jason McC. Smith and David Stotts. SPQR: Flexible automated design pattern extraction from source code. In 18th IEEE International Conference on Automated Software Engineering, pp. 215–224, Oct 2003.
[38] Jason McC. Smith 和 David Stotts。使用 SPQR 进行面向意图的设计模式形式化，第 7 章。IDEA Group，2007 年。
[38] Jason McC. Smith and David Stotts. Intent-Oriented Design Pattern Formalization Using SPQR, chapter 7. IDEA Group, 2007.
[39] 马克·吐温。《汤姆索亚历险记》。古腾堡出版社，www.gutenberg.org/ ebooks/74，2005年。
[39] Mark Twain. The Adventures of Tom Sawyer. Gutenberg Press, www.gutenberg.org/ebooks/74, 2005.
[40] Bobby Woolf。抽象类模式。Neil Harrison、Brian Foote 和 Hans Rohnert 编，《程序设计模式语言 4》。Addison-Wesley，波士顿，1998 年。
[40] Bobby Woolf. The abstract class pattern. In Neil Harrison, Brian Foote, and Hans Rohnert, eds., Pattern Languages of Program Design 4. Addison-Wesley, Boston, 1998.
[41] Bobby Woolf。对象递归模式。Neil Harrison、Brian Foote 和 Hans Rohnert 编，《程序设计模式语言 4》。Addison-Wesley，波士顿，1998 年，第 41-52 页。
[41] Bobby Woolf. The object recursion pattern. In Neil Harrison, Brian Foote, and Hans Rohnert, eds., Pattern Languages of Program Design 4. Addison-Wesley, Boston, 1998, 41–52.
[42] E. Yourdon 和 L. Constantine。结构化设计。Prentice Hall，新泽西州恩格尔伍德克利夫斯，1979 年。
[42] E. Yourdon and L. Constantine. Structured Design. Prentice Hall, Englewood Cliffs, NJ, 1979.
[43] Walter Zimmer。设计模式之间的关系。James O. Coplien 和 Douglas C. Schmidt 编，《程序设计的模式语言》。Addison-Wesley，波士顿，1995 年，第 345-364 页。
[43] Walter Zimmer. Relationships between design patterns. In James O. Coplien and Douglas C. Schmidt, eds., Pattern Languages of Program Design. Addison-Wesley, Boston, 1995, 345–364.
指数
Index
一个
A
阿巴迪·马丁，281
Abadi, Martín, 281
抽象工厂模式
Abstract Factory pattern
作为合并的PINbox，65
as coalesced PINbox, 65
折叠 PINbox，50
as collapsed PINbox, 50
作为扩展的 PINbox，64
as expanded PINbox, 64
作为堆叠 PINbox，60
as stacked PINbox, 60
作为 UML，59
as UML, 59
rho-演算，313
rho-calculus, 313
作为Fulfill Method模式的一部分，68 – 72
as part of Fulfill Method pattern, 68–72
合作，142
collaborations, 142
意图，140
intent, 140
同位素形式，75
isotopic forms, 75
参与者，142
participants, 142
相关模式，144
related patterns, 144
rho-演算，298
rho-calculus, 298
结构，142
structure, 142
Abstraction density, EDPs, 110–112
阿克曼，李，9岁
Ackerman, Lee, 9
Alexander, Christopher, 2, 4, 86
动物示例
Animal example
吃。请参阅对象化模式。
eating. See Objectifier pattern.
正在移动。请参阅抽象接口模式。
moving. See Abstract Interface pattern.
反模式，91
Anti-patterns, 91
数组排序示例。请参阅 递归模式。
Array sorting example. See Recursion pattern.
异步执行，146
Asynchronous execution, 146
乙
B
布奇，格雷迪
Booch, Grady
图案集锦，4
pattern collections, 4
书籍和出版物
Books and publications
清洁代码, 32
Clean Code, 32
设计模式：元素...，2
Design Patterns: Elements of..., 2
重构, 92
Refactoring, 92
重构模式, 92
Refactoring to Patterns, 92
碳
C
C代码示例，118、120、121、193、246
C code examples, 118, 120, 121, 193, 246
C# 代码示例，178
C# code examples, 178
C ++代码示例、26、29、31、36、125、128、138、143、148、153、160、166 – 168、173、176、184、190、193 – 195、201、210、213、216、220、222、227、233、238、247
C++ code examples, 26, 29, 31, 36, 125, 128, 138, 143, 148, 153, 160, 166–168, 173, 176, 184, 190, 193–195, 201, 210, 213, 216, 220, 222, 227, 233, 238, 247
调用。请参阅 委托模式。
Calls. See Delegation pattern.
卡德利，卢卡，281
Cardelli, Luca, 281
模式定义的类别，22
Categories of pattern definition, 22
CEO example, 36. See also Delegation pattern.
责任链模式
Chain of Responsibility pattern
rho-演算，318
rho-calculus, 318
柴郡猫技术，119
Cheshire cat technique, 119
Clarity of expression, EDPs, 109–110
课程
Classes
objectifying similar behaviors, 244. See also Objectifier pattern.
清洁代码, 32
Clean Code, 32
凝聚
Cohesion
对象，35
objects, 35
通用 Lisp 对象系统 (CLOS)，10
Common Lisp Object System (CLOS), 10
Composition, rho-calculus, 291–293
聚集模式。另请参阅 委托聚集模式。
Conglomeration pattern. See also Delegated Conglomeration pattern.
适用性，161
applicability, 161
as part of Factory Method pattern, 263–265
as part of Template pattern, 275–279
合作，162
collaborations, 162
后果，162
consequences, 162
实施，162
implementation, 162
意图，159
intent, 159
method call classification, 163–164
rho-微积分，300
rho-calculus, 300
结构，161
structure, 161
上下文类别, 22
Context category, 22
适用性，122
applicability, 122
as part of Abstract Factory pattern, 261–263
as part of Factory Method pattern, 263–265
合作，123
collaborations, 123
后果，124
consequences, 124
意图，117
intent, 117
参与者，123
participants, 123
相关模式，125
related patterns, 125
结构，123
structure, 123
德
D
数据协议回退示例。请参阅 恢复方法模式。
Data protocol fallback example. See Revert Method pattern.
分解消息。请参阅 聚合模式。
Decomposing Message. See Conglomeration pattern.
模式定义的类别, 22
categories of pattern definition, 22
上下文类别, 22
context category, 22
example, Decorator pattern, 18–21
目标，22
goal of, 22
面向对象编程，实体
object-oriented programming, entity
互动，28
interactions, 28
问题类别, 22
problem category, 22
解决方案类别, 22
solution category, 22
未命名的命名空间, 24
unnamed namespaces, 24
装饰器模式
Decorator pattern
作为协作 UML 元素, 47
as collaboration UML element, 47
作为扩展的 PINbox，57
as expanded PINbox, 57
作为 PINbox，52
as PINbox, 52
rho-演算，316
rho-calculus, 316
装饰器模式，示例
Decorator pattern, examples
推迟实现模式。请参阅 抽象接口模式。
Defer Implementation pattern. See Abstract Interface pattern.
委托集聚模式。另请参阅 集聚模式；委托模式；代理委托模式；受信任委托模式。
Delegated Conglomeration pattern. See also Conglomeration pattern; Delegation pattern; Deputized Delegation pattern; Trusted Delegation pattern.
适用性，189
applicability, 189
合作，190
collaborations, 190
后果，190
consequences, 190
意图，187
intent, 187
method call classification, 191–192
相关模式，191
related patterns, 191
rho-演算，303
rho-calculus, 303
结构，189
structure, 189
委托模式。另请参阅 委托聚集模式；代理委托模式；受信任委托模式。
Delegation pattern. See also Delegated Conglomeration pattern; Deputized Delegation pattern; Trusted Delegation pattern.
适用性，147
applicability, 147
合作，148
collaborations, 148
后果，148
consequences, 148
执行情况，148
implementation, 148
意图，145
intent, 145
method call classification, 149–150
参与者，147
participants, 147
相关模式，149
related patterns, 149
rho 演算, 299
rho-calculus, 299
结构，147
structure, 147
Delegated Conglomeration, 187–192
代理委托模式。另请参阅 代理聚集模式；委托模式；受信任委托模式。
Deputized Delegation pattern. See also Delegated Conglomeration pattern; Delegation pattern; Trusted Delegation pattern.
适用性，218
applicability, 218
合作，219
collaborations, 219
实施，220
implementation, 220
意图，216
intent, 216
method call classification, 220–221
相关模式，220
related patterns, 220
rho-演算，306
rho-calculus, 306
结构，219
structure, 219
代理重定向模式。另请参阅 重定向递归模式；重定向模式；受信任重定向模式。
Deputized Redirection pattern. See also Redirected Recursion pattern; Redirection pattern; Trusted Redirection pattern.
适用性，224
applicability, 224
as part of Proxy pattern, 269–273
合作，226
collaborations, 226
后果，227
consequences, 227
实施，227
implementation, 227
意图，222
intent, 222
method call classification, 227–228
相关模式，227
related patterns, 227
rho-演算，307
rho-calculus, 307
结构，226
structure, 226
设计模式。另请参阅 EDP（元素设计模式）。
Design patterns. See also EDPs (Elemental Design Patterns).
反模式, 91
anti-patterns, 91
合集，4
collections of, 4
定义，21
definition, 21
图表。参见 PIN（模式实例符号）；PIN 盒。
diagramming. See PIN (Pattern Instance Notation); PINboxes.
医源性，91
iatrogenic, 91
instance notation and roles, 293–295
恶性，91
malignant, 91
部分, 91
partial, 91
设计模式，历史
Design patterns, history of
language-dependent views, 10–12
从神话到科学，12
from myth to science, 12
图案集锦，4
pattern collections, 4
开创性的作品，2
seminal works, 2
设计模式：元素...，2
Design Patterns: Elements of..., 2
绘制设计模式图。参见 PIN（模式实例符号）；PINbox。
Diagramming design patterns. See PIN (Pattern Instance Notation); PINboxes.
文档
Documentation
history of design patterns, 5–8
点运算符，283
Dot operators, 283
双调度，10
Double-dispatch, 10
d 指针技术, 119
d-pointer technique, 119
埃
E
EDP（基本设计模式）。另请参阅 设计模式。
EDPs (Elemental Design Patterns). See also Design patterns.
反模式, 91
anti-patterns, 91
分解成更小的部分。请参阅 分解模式。
breaking into smaller parts. See Decomposing patterns.
clarity of expression, 109–110
combining with other patterns, 101–105
概念与结构，16
concepts vs. constructs, 16
定义，18
definition, 18
医源性模式，91
iatrogenic patterns, 91
恶性模式，91
malignant patterns, 91
方法相似性, 32
method similarity, 32
部分设计模式，91
partial design patterns, 91
程序分析，112
procedural analysis, 112
recreating Decorator, example, 77–90
rho-演算。参见 Rho-演算、EDP 定义。
rho-calculus. See Rho-calculus, EDP definitions.
SPQR (System for Pattern Query and Recognition), 14–17
used-by relationships, 101–102
EDP（元素设计模式）、对象
EDPs (Elemental Design Patterns), objects
Abstract Interface pattern, 43–44
凝聚力，35
cohesion, 35
继承模式, 43
Inheritance pattern, 43
封装
Encapsulation
相似性，32
similarity, 32
数据。请参阅 创建对象模式。
of data. See Create Object pattern.
设计，101
of design, 101
执行模式。请参阅 委托模式。
Executive pattern. See Delegation pattern.
适用性，183
applicability, 183
as part of Decorator pattern, 83–86, 268–269
as part of Chain of Responsibility pattern, 273–275
合作，184
collaborations, 184
后果，184
consequences, 184
意图，181
intent, 181
method call classification, 185–186
参与者，183
participants, 183
相关模式，185
related patterns, 185
rho-演算，302
rho-calculus, 302
结构，183
structure, 183
扩展超级模式。请参阅 扩展方法模式。
Extending Super pattern. See Extend Method pattern.
F
F
Factory Method pattern description, 263–265
Fields, decomposing patterns, 25–30
享元模式作为PINbox，53
Flyweight pattern as PINbox, 53
Formalism of patterns, 105–108. See also Rho-calculus.
Founders, history of design patterns, 2, 4. See also specific names.
福勒·马丁，92岁
Fowler, Martin, 92
适用性，231
applicability, 231
as part of Abstract Factory pattern, 261–263
as part of Factory Method pattern, 263–265
作为Objectifier模式的一部分，77、79 – 80、248 – 249
as part of Objectifier pattern, 77, 79–80, 248–249
as part of Proxy pattern, 269–273
as part of Template pattern, 275–279
合作，232
collaborations, 232
后果，233
consequences, 233
意图，231
intent, 231
动机，231
motivation, 231
参与者，231
participants, 231
相关模式，234
related patterns, 234
rho-演算，308
rho-calculus, 308
结构，232
structure, 232
格
G
伽马，埃里希，2
Gamma, Erich, 2
四人帮模式
Gang of Four patterns
Abstract Factory, 260–263, 313
Chain of Responsibility, 273–275, 318
Factory Method, 263–265, 314–315
赫
H
赫尔姆，理查德，2
Helm, Richard, 2
辅助方法。请参阅 聚集模式。
Helper Methods. See Conglomeration pattern.
设计模式的历史
History of design patterns
founders, 2, 4. See also specific names.
language-dependent views, 10–12
从神话到科学，12
from myth to science, 12
图案集锦，4
pattern collections, 4
开创性的作品，2
seminal works, 2
我
I
继承，多重，136
Inheritance, multiple, 136
适用性，135
applicability, 135
作为Fulfill Method模式的一部分，68 – 72
as part of Fulfill Method pattern, 68–72
合作，135
collaborations, 135
创建对象, 43
creating objects, 43
意图，130
intent, 130
同位素形式，76
isotopic forms, 76
参与者，135
participants, 135
相关模式, 138
related patterns, 138
rho-演算，298
rho-calculus, 298
结构，135
structure, 135
Instances, creating, 235–239. See also Retrieve New pattern.
实例化模式。请参阅 创建对象模式。
Instantiation pattern. See Create Object pattern.
IsA模式。请参阅 继承模式。
IsA pattern. See Inheritance pattern.
J
J
Java代码示例、27、36、38、126、127、129、133 – 134、141、144、145、152、156、162、169、184、187、188、197、206、233、243
Java code examples, 27, 36, 38, 126, 127, 129, 133–134, 141, 144, 145, 152, 156, 162, 169, 184, 187, 188, 197, 206, 233, 243
约翰逊，拉尔夫，2
Johnson, Ralph, 2
钾
K
乔舒亚·克里夫斯基（Joshua Kerievsky），92岁
Kerievsky, Joshua, 92
大号
L
语言习语，11
Language idioms, 11
Language-dependent patterns, 10–12
米
M
恶性模式，91
Malignant patterns, 91
马丁·罗伯特32 岁
Martin, Robert, 32
Mathematics of patterns, 105–108. See also Rho-calculus.
内存管理。请参阅对象管理。
Memory management. See Object management.
消息传递模式。请参阅 委托模式。
Messaging pattern. See Delegation pattern.
方法调用模式。请参阅 委托模式。
Method Invocation pattern. See Delegation pattern.
方法调用模式
Method Invocation patterns
Delegated Conglomeration, 187–192
Deputized Redirection, 222–228
方法
Methods
超载，32
overloaded, 32
先前提取的实现，231 – 234。另请参阅 Fulfill Method模式。
previously extracted, implementing, 231–234. See also Fulfill Method pattern.
相似性，32
similarity, 32
莫斯示例，7
Moss example, 7
迈耶斯·斯科特，124
Meyers, Scott, 124
Multiplicity connections, PINboxes, 56–62
否
N
命名空间，未命名，24
Namespaces, unnamed, 24
命名约定, 18
Naming conventions, 18
哦
O
对象元素模式
Object Elements patterns
C++ shared_ptr，240
C++ shared_ptr, 240
C++ 非托管，235
C++ unmanaged, 235
垃圾收集，235，238，240，243
garbage collection, 235, 238, 240, 243
Objective-C 自动释放, 240
Objective-C autorelease, 240
Object Recursion pattern. See also Recursion pattern; Redirected Recursion pattern.
as part of Decorator pattern, 83–86, 268–269
后果，256
consequences, 256
继承，133
in Inheritance, 133
意图，251
intent, 251
参与者，253
participants, 253
相关模式，257
related patterns, 257
rho-演算，312
rho-calculus, 312
合作，249
collaborations, 249
执行情况，250
implementation, 250
意图，244
intent, 244
参与者，249
participants, 249
相关模式，250
related patterns, 250
rho-演算，311
rho-calculus, 311
Objective - C代码示例，37、131、132、137、156、190
Objective-C code examples, 37, 131, 132, 137, 156, 190
面向对象编程，实体交互，28
Object-oriented programming, entity interactions, 28
对象
Objects
Abstract Interface pattern, 43–44
凝聚力，35
cohesion, 35
Create Object pattern, 117–125
隐藏细节。请参阅 代理模式。
hiding details. See Proxy pattern.
继承模式, 43
Inheritance pattern, 43
限制访问。请参阅 代理模式。
limiting access. See Proxy pattern.
共享，引用但不拥有，240 – 243。另请参阅 检索共享模式。
shared, referencing without owning, 240–243. See also Retrieve Shared pattern.
不透明指针技术，119
Opaque pointer technique, 119
重写方法
Overriding a method
继承，133
in Inheritance, 133
需要。请参阅抽象接口模式。
requiring. See Abstract Interface pattern.
无需替换。请参阅 扩展方法模式。
without replacing. See Extend Method pattern.
磷
P
Painter example. See Redirection pattern, Conglomeration pattern.
伞兵示例。请参阅 重定向递归模式。
Paratroopers example. See Redirected Recursion pattern.
部分图案，91
Partial patterns, 91
模式。请参阅 设计模式。
Patterns. See Design patterns.
PIN (Pattern Instance Notation), 45–49
定义，49
definition, 49
指向实现的指针（pimpl）技术，119
Pointer-to-implementation (pimpl) technique, 119
Polymorphism. See Abstract Interface pattern; Object Recursion pattern.
多态模式。请参阅 抽象接口模式。
Polymorphism pattern. See Abstract Interface pattern.
问题类别，22
Problem category, 22
程序分析，EDP，112
Procedural analysis, EDPs, 112
rho-演算，317
rho-calculus, 317
Python代码示例，129、138、144、162、182、197、206、233、238
Python code examples, 129, 138, 144, 162, 182, 197, 206, 233, 238
R
R
递归，定义，165
Recursion, definition, 165
Recursion pattern. See also Object Recursion pattern; Redirected Recursion pattern.
适用性，168
applicability, 168
合作，169
collaborations, 169
后果，169
consequences, 169
实施，169
implementation, 169
意图，165
intent, 165
method call classification, 170–171
rho-演算，301
rho-calculus, 301
结构，169
structure, 169
重定向递归模式。另请参阅 代理重定向模式；对象递归模式；递归模式；重定向模式；受信任重定向模式。
Redirected Recursion pattern. See also Deputized Redirection pattern; Object Recursion pattern; Recursion pattern; Redirection pattern; Trusted Redirection pattern.
适用性，196
applicability, 196
as part of Chain of Responsibility pattern, 273–275
合作，197
collaborations, 197
后果，197
consequences, 197
设计空间，40
design space, 40
意图，193
intent, 193
method call classification, 198–199
参与者，197
participants, 197
相关模式, 198
related patterns, 198
rho-演算，303
rho-calculus, 303
结构，196
structure, 196
重定向模式。另请参阅 代理重定向模式；重定向递归模式；受信任重定向模式。
Redirection pattern. See also Deputized Redirection pattern; Redirected Recursion pattern; Trusted Redirection pattern.
合作，155
collaborations, 155
后果，155
consequences, 155
执行情况，156
implementation, 156
意图，151
intent, 151
method call classification, 157–158
参与者，154
participants, 154
相关模式，157
related patterns, 157
rho-微积分，300
rho-calculus, 300
结构，154
structure, 154
重定向模式
Redirection patterns
Deputized Redirection, 222–228
Reduction rules, 285–286, 291–293
重构, 92
Refactoring, 92
重构模式, 92
Refactoring to Patterns, 92
重构
Refactorings
内审员之间，106
between EDPs, 106
请求，分布式处理。请参阅 对象递归模式。
Requests, distributed handling. See Object Recursion pattern.
适用性，236
applicability, 236
后果，238
consequences, 238
意图，235
intent, 235
参与者，237
participants, 237
相关模式，239
related patterns, 239
rho-演算，309
rho-calculus, 309
检索模式
Retrieve pattern
适用性，127
applicability, 127
as part of Abstract Factory pattern, 261–263
as part of Factory Method pattern, 263–265
合作，128
collaborations, 128
后果，128
consequences, 128
意图，126
intent, 126
参与者，128
participants, 128
相关模式，129
related patterns, 129
结构，127
structure, 127
适用性，241
applicability, 241
合作，243
collaborations, 243
后果，243
consequences, 243
实施，243
implementation, 243
意图，240
intent, 240
参与者，242
participants, 242
相关模式，243
related patterns, 243
rho-演算，310
rho-calculus, 310
适用性，177
applicability, 177
合作，178
collaborations, 178
后果，178
consequences, 178
意图，172
intent, 172
method call classification, 179–180
相关模式，179
related patterns, 179
rho-演算，301
rho-calculus, 301
结构，177
structure, 177
点运算符，283
dot operators, 283
pattern instance notation and roles, 293–295
reduction rules, 285–286, 291–293
范围界定, 283
scoping, 283
抽象接口, 298
Abstract Interface, 298
集团, 300
Conglomeration, 300
委托集团, 303
Delegated Conglomeration, 303
代表团，299
Delegation, 299
代表代表团, 306
Deputized Delegation, 306
代理重定向, 307
Deputized Redirection, 307
扩展方法, 302
Extend Method, 302
继承，298
Inheritance, 298
递归，301
Recursion, 301
重定向递归, 303
Redirected Recursion, 303
重定向，300
Redirection, 300
还原方法, 301
Revert Method, 301
受信任的委托, 304
Trusted Delegation, 304
受信任的重定向, 305
Trusted Redirection, 305
Rho-演算，四人帮定义
Rho-calculus, Gang of Four definitions
抽象工厂, 313
Abstract Factory, 313
责任链, 318
Chain of Responsibility, 318
装饰者，316
Decorator, 316
代理，317
Proxy, 317
模板方法, 319
Template Method, 319
Rho-演算，中级定义
Rho-calculus, intermediate definitions
履行方法, 308
Fulfill Method, 308
对象递归, 312
Object Recursion, 312
客体化者, 311
Objectifier, 311
检索新, 309
Retrieve New, 309
检索共享, 310
Retrieve Shared, 310
年代
S
车间领班。请参阅 重定向模式。
Shop foreman. See Redirection pattern.
相似
Similarity
对象，32
objects, 32
单例模式
Singleton pattern
折叠 PINbox，50
as collapsed PINbox, 50
社交网络邀请示例。请参阅 委托聚合模式。
Social network invites example. See Delegated Conglomeration pattern.
解决方案类别, 22
Solution category, 22
SPQR (System for Pattern Query and Recognition), 14–17
策略模式
Strategy pattern
作为 UML 中的 pattern:role 标签，48
as pattern:role tags in UML, 48
UML 中的多个实例作为模式：角色标签, 49
multiple instances as pattern:role tags in UML, 49
multiple instances as PINboxes, 61, 62
同步方法调用，146
Synchronous method calls, 146
电视
T
作为工厂方法模式的一部分，279
as part of Factory Method pattern, 279
序列图, 51
in sequence diagram, 51
rho-演算，319
rho-calculus, 319
汤姆索亚。请参阅 重定向模式。
Tom Sawyer. See Redirection pattern.
Transitivity, rho-calculus, 285–286
适用性，204
applicability, 204
合作，205
collaborations, 205
后果，205
consequences, 205
意图，200
intent, 200
method call classification, 207–208
参与者，205
participants, 205
rho-演算，304
rho-calculus, 304
结构，204
structure, 204
受信任的重定向模式。另请参阅 代理重定向模式；重定向递归模式；重定向模式。
Trusted Redirection pattern. See also Deputized Redirection pattern; Redirected Recursion pattern; Redirection pattern.
适用性，212
applicability, 212
作为对象递归模式的一部分，79、81 – 83、254 – 256
as part of Object Recursion pattern, 79, 81–83, 254–256
合作，213
collaborations, 213
后果，213
consequences, 213
实施，213
implementation, 213
意图，209
intent, 209
method call classification, 214–215
rho-演算，305
rho-calculus, 305
结构，212
structure, 212
类型关系模式
Type Relation patterns
类型重用模式。请参阅 继承模式。
Type Reuse pattern. See Inheritance pattern.
Types, decomposing patterns, 25–30
乌
U
UML（统一建模语言），46 – 49。另请参阅 PIN （模式实例符号）；PINbox。
UML (Unified Modeling Language), 46–49. See also PIN (Pattern Instance Notation); PINboxes.
未命名的命名空间，24
Unnamed namespaces, 24
Used-by relationships, EDPs, 101–102
五
V
虚拟方法模式。请参阅 抽象接口模式。
Virtual Method pattern. See Abstract Interface pattern.
访客模式，10
Visitor pattern, 10
Vlissides，约翰，2
Vlissides, John, 2
西
W
是
Z
齐默，沃尔特，244
Zimmer, Walter, 244
脚注
Footnotes
第 1 章
Chapter 1
1.您现在不需要知道访问者模式是什么。我之所以选择它，只是因为对访问者模式的讨论明确地解决了我的观点。
1. You don’t need to know what the Visitor pattern is right now. I selected it only because the discussion of Visitor explicitly addresses the point I’m making.
2.有些设计模式是特定语言所独有的，而且只存在于这些语言中，但这些模式通常被称为语言习语。在本文中，当我们使用术语“设计模式”时，我们特指与语言无关的概念。
2. Some design patterns are unique to specific languages, and only those languages, but those patterns are often called language idioms. In this text when we use the term design patterns, we are specifically talking about concepts that are language independent.
第 2 章
Chapter 2
1.必须完全一样吗？不！我们稍后会讲到。
1. Does it have to be exactly the same? No! We’ll get to that in a moment.
2.我们本可以使用递归重定向，但在早期的反馈中，该术语对大多数人来说意味着两个对象之间的循环，而出于某种原因，这个术语并不如此。此外，如果您有兴趣深入了解附录，还会给出正式的理由。
2. We could have gone with Recursive Redirection, but in early feedback, that term implied to most people a loop between two objects, while this one did not, for some reason. Also, there’s a formal reason given if you care to dive in to the appendix.
3.如果您不熟悉多态性这个术语，请从继承EDP 开始。这将让您很好地了解这个概念。
3. If you’re unfamiliar with the term polymorphism, start with the Inheritance EDP. That will give you a good overview of the concept.
4.回想一下，依赖关系从调用方法转到被调用的方法；在这种情况下，调用方法是在包含被调用方法的类型的子类型中定义的。
4. Recall that reliances go from the calling method to the method that is called; in this case the calling method is defined within a subtype of the type enclosing the method being called.
第四章
Chapter 4
1.这并不完全正确：有些语言确实允许通过后续抽象来删除方法，但我们不会在这里讨论它们，因为这是一个高级理论技巧，远远超出了本书的范围。此外，这样的删除将是一个非常不同的概念，不是吗？我们不会履行先前承诺的方法实现，我们会删除对先前方法实现的访问权限。
1. This isn’t strictly true: some languages do allow for such removal of methods via later abstraction, but we won’t get into them here because that’s an advanced theoretical trick well beyond the scope of this book. Besides, such a removal would be a very different concept, wouldn’t it? We wouldn’t be fulfilling a prior promised method implementation, we’d be erasing access to prior method implementations.
2.如果角色确实不同，则表明设计模式发生了根本性变化。这种变化称为模式的变体，表明某些非常重大的变化已经发生，例如正在解决的问题或发生问题的上下文。
2. If the roles do differ, it indicates a fundamental change to the design pattern. This change is called a variant of the pattern and is a clue that something very substantial has changed, such as the problem being solved or the context in which it is occurring.
3.这是传递性的另一个例子，它是附录中描述的ρ-演算的核心特征，也是EDP真正强大的原因。你看，我告诉过你，我会不断提醒你去阅读正式的部分。
3. This is another example of transitivity, and it is a core feature of the ρ-calculus described in the appendix, and what gives the EDPs their real power. See, I told you I was going to keep giving you nudges to go read the formal bits.
4. ConcreteDecoratorA并Decorator形成继承的另一支柱，但为了清晰起见，我们暂时将其排除在外。
4. ConcreteDecoratorA and Decorator form another leg of Inheritance, but we’re leaving it out for clarity at the moment.
5.再次，我们暂时忽略重载。假设相似性不仅基于名称，还基于整个方法签名。
5. Once again, we’re ignoring overloading for now. Assume that similarity is based not only on the name but on the entire method signature.
第五章
Chapter 5
1.这是早期 C++ 系统的内部结构，当时该cfront工具是编译 C++ 代码的主要技术。如果您对面向对象语言和过程语言之间的界限感兴趣，那么这是一个很好的历史起点。
1. This is what early C++ systems looked like under the hood, when the cfront tool was the primary technique for compiling C++ code. If you’re interested in the boundary between object-oriented and procedural languages, that is a good historical place to start.
2.有一种流行的语言是这条规则的一个明显例外：Objective-C 不像大多数语言那样在一个行为中同时强制分配和初始化。相反，它有一个 alloc-init 习语，旨在让开发人员将对象的内存管理和数据初始化作为单独的部分来处理。虽然它可以提供优化机会（正如您所预料的），但它也可能导致开发人员犯错。
2. One popular language is a glaring exception to this rule: Objective-C does not enforce both allocation and initialization in one behavior, as most languages do. Instead, it has an alloc-init idiom intended to let developers handle the memory management and data initialization of objects as separate pieces. While it can offer optimization opportunities, as you might expect, it can also lead to developer error.
3.顺便说一句，调用超类版本的相同方法是扩展方法EDP 的一个示例。您可以阅读该条目以了解更深入的讨论。
3. By the way, this calling of the superclass’s version of the same method is an example of the Extend Method EDP. You can read that entry for a more thorough discussion.
4.这也让我有了我最喜欢的个性车牌之一。那是一辆 60 年代中期的大众甲壳虫，停在西雅图，车牌上写着FEATURE。因为任何 bug，如果存在的时间足够长，最终都会成为某个地方某个人所依赖的功能。
4. It also resulted in one of my favorite vanity plates. It was a mid-60s VW Beetle in Seattle with the license plate FEATURE. Because any bug, if left long enough, eventually becomes a feature that someone, somewhere, is relying on.
5.你甚至可以说，就接口而言，继承可以被认为是多个重定向模式实例的聚集，Redirect它们之间共享一个对象。
5. You can even say that, as far as the interface is concerned, Inheritance can be considered a gathering together of multiple Redirection pattern instances, with one Redirect object being shared among them.
6. C++ 在函子概念中提供了类似的东西，用作operator()调用机制，但 C# 更进一步，允许临时包装预先存在的方法，以便以这种方式传递。
6. C++ offers something similar in the functor concept using operator() as the calling mechanism, but C# takes the next step to allow for ad hoc wrapping of preexisting methods for passing around in this manner.
7.有一个很好的理由可以证明，让文本输入小部件控制数字显示确实具有合理的设计意义，假设数字输入控件仅限于数字，但有些 UI 工具包却采用了另一种方式。
7. A good argument can be made that having a text input widget control a numeric display does make rational design sense, assuming a numeric input control that was restricted to numerics only, but there are UI toolkits that went the other way.
附录 A
Appendix A
1.除了一些例外，尤其是 Eiffel；这里就不详细介绍了，但在 [ 35 ] 中进行了介绍。如果你想进一步了解，可以先研究一下方法的协变和逆变返回类型。
1. With some exceptions, notably Eiffel; the details are too much for here, but are covered in [35]. If you wish to dive further, investigate covariant versus contravariant return types of methods as a starting point.
2.是的，这似乎有点不拘泥于类型。我建议你参考ς-calculus [ 1 ] 的文本，它应该很快就能告诉你事实并非如此。基础是牢固的。
2. Yes, this appears to be playing a bit fast and loose with types. I refer you to the text for ς-calculus [1], which should quickly make clear that this is not the case. The foundation is sound.
3.从形式上看，子类型、子类化和继承并不完全相同，但就本书而言，我们可以互换使用这些术语。详情请参阅 [ 1 ]。
3. Subtyping, subclassing, and inheritance are not all the same from a formal point of view, but for the purposes of this book, we can use the terms interchangeably. See [1] for details.
4. ~ 和≁符号（请原谅我的双关语）太过相似，下标字体太小，因此无法以这种方式使用。根据用户反馈，我们采用了 +/– 作为更明显的版本。
4. The ~ and ≁ notations were, pardon the pun, too similar at the small font sizes found in subscripts to be useful in this manner. User feedback led to the adoption of the +/– as a more visible version.